How do changes in pension assumptions inflate or deflate reported earnings?

Pension accounting is one of the most opaque corners of financial reporting. A company with a large, underfunded pension plan has leverage to manipulate reported earnings through seemingly minor adjustments to assumptions. By lowering the discount rate used to calculate pension liabilities, raising the assumed return on plan assets, or tweaking mortality or salary growth assumptions, finance teams can reduce pension expense in the income statement and boost reported earnings.

These moves are legal under GAAP. They are not fraud. But they are discretionary, and they mask underlying pension funding weakness. An investor who ignores pension assumptions misses one of the most effective, least scrutinized levers for earnings management in corporate America.

This article teaches you to decode the pension footnote, spot unreasonable assumptions, and adjust reported earnings to account for discretionary pension choices.

Quick definition: A pension is a defined-benefit obligation to pay retired employees a stream of payments. Companies recognize pension expense on the income statement based on assumptions about discount rates, asset returns, salary growth, and mortality. Adjusting these assumptions can significantly reduce or increase reported expense—and earnings—without changing the actual pension obligation.

Key takeaways

• Pension expense appears on the income statement and is influenced by three main drivers: the discount rate (used to value liabilities), the assumed return on plan assets (offset against expense), and demographic assumptions (salary growth, mortality).
• A lower discount rate increases the present value of pension liabilities, raising expense. A higher assumed asset return decreases expense. Companies can adjust these within GAAP, making pension accounting subjective.
• Changes in discount rates are partly external (they follow bond yields) but are applied with judgment. Companies can cherry-pick the measurement date or settlement convention to favor lower rates.
• The assumed return on plan assets is purely discretionary and is a common manipulation target. A company might assume a 7% return even if historical returns are 5%, inflating earnings.
• Pension gains and losses from market movements are recognized in other comprehensive income (OCI), creating a timing mismatch. This can mask deteriorating funded status.
• Forensic investors compare a company's pension assumptions to peer assumptions and to historical returns to flag unreasonable choices.

What is pension expense and where does it come from?

A defined-benefit pension plan is a promise to pay a retired employee a stream of income, typically based on tenure, salary, and age. The company funds the plan during the employee's working years and invests the assets, hoping to earn enough to cover future payments.

Pension expense on the income statement includes several components:

• Service cost: The increase in the pension obligation due to one year of additional service. This is relatively straightforward to calculate and is the most defensible part of pension expense.
• Interest cost: The accrual of interest on the existing obligation, based on the discount rate applied to the pension liability.
• Expected return on assets: A benefit (negative expense) based on the assumed return on plan assets. The actual return earned by the plan is deferred and recognized later, in OCI.
• Amortization of deferred gains and losses: Prior-year gains or losses (from actual returns differing from expected returns, or assumption changes) are amortized over time, smoothing the impact on the income statement.

The formula (simplified):

Pension Expense = Service Cost + Interest Cost – Expected Return on Assets + Amortization of Deferred Gains/Losses

Each of these components involves judgment, but the discount rate and expected return on assets are the two biggest levers.

The discount rate: external pressure, internal discretion

The discount rate is used to calculate the present value of the pension obligation. A lower rate makes the obligation larger; a higher rate makes it smaller. The discount rate should reflect the yield on high-quality corporate bonds (typically, the yield on long-duration investment-grade bonds matching the duration of the obligation).

In theory, the discount rate is driven by market yields, so companies have little discretion. In practice, they do:

• Measurement date timing. Companies can measure the obligation on a day when yields are lower (or higher), affecting the rate used. A company that measures on December 31st can face different bond yields than one measuring on January 15th. Over many years, this timing bias can add up.
• Discount rate construction. Companies use different bond yield curves (AA, BBB, Treasury plus spreads). Different curves yield different rates. A choice that appears technical can swing the discount rate 50–100 basis points.
• Liability duration matching. Some plans use a "settlement approach," matching discount rates to the duration of expected payouts. Others use a simpler single-rate approach. The settlement method typically produces lower rates (higher liabilities) but some companies resist this transparency.

A company that has lowered its discount rate year over year is worth investigating. If the rate decline exceeds changes in market yields, suspect discretion.

The assumed return on assets: the biggest discretionary lever

The assumed return on plan assets is the expected long-term return the company expects its pension assets to earn. It is used to calculate the "expected return on assets" benefit that offsets pension expense.

Example: A pension plan has $1 billion in assets. The company assumes a 7% annual return. The expected return on assets is $70 million, which reduces pension expense by that amount.

The problem: this expected return is an assumption, not reality. If the plan actually earns 5%, the shortfall (2% × $1 billion = $20 million) is deferred and not immediately expensed. It is amortized in future years as a "deferred loss." The timing mismatch allows companies to smooth earnings over many years.

More importantly, the assumed return is purely discretionary. A company can assume 7% even if historical returns are 5%, if peers assume 6%, or if the yield environment makes 7% unrealistic. GAAP does not mandate a specific assumption; it requires only that the assumption be "reasonable" in light of historical returns and expectations. Reasonableness is in the eye of the beholder.

A company with an assumed return higher than peers, or higher than the company's own historical returns, is using the assumption to depress reported pension expense. Over time, this creates a "deferred loss" on the balance sheet (in AOCI), which eventually must reverse and hit earnings.

How companies exploit pension assumptions to manage earnings

The playbook is simple and effective:

Year 1–2: Times are good. The company reports solid earnings. Management leaves pension assumptions at conservative levels (e.g., 6.5% assumed return, 4.2% discount rate).

Year 3–5: Business slows. Earnings pressure mounts. Management quietly raises the assumed return to 7% (from 6.5%) and lowers the discount rate to 4.0% (from 4.2%). The change looks minor on the surface.

Impact: Raising the assumed return from 6.5% to 7% on a $2 billion plan reduces pension expense by $30 million (0.5% × $2B). Lowering the discount rate increases interest cost, which should increase expense, but management can offset this by adjusting other assumptions.

Net effect: Pension expense shrinks by $20–30 million per year, just as operating earnings are declining. Reported total earnings are propped up.

Years 6–7: The plan faces market volatility or underperforms assumptions. Deferred losses accumulate. Management must eventually amortize these losses, hitting future earnings. But by then, the executive team may have changed, or attention has moved to other issues.

The discount rate: when it is out of line

To spot a suspicious discount rate, compare three things:

1. The company's current rate to its prior-year rate. A shift of more than 50 basis points without corresponding market yield changes is suspect.
2. The company's current rate to peers. If the industry average discount rate is 4.5% and your company uses 4.0%, ask why. Tax-exempt plans (for universities and nonprofits) use different rates, but for-profits in the same industry should be roughly aligned.
3. The company's current rate to relevant bond yields. The discount rate should roughly match the yield on high-quality corporate bonds with duration similar to the plan's obligations. If 10-year BBB yields are 5.0% and the company uses 4.2%, the company is choosing a lower rate to minimize liabilities.

A company that shifts its discount rate downward while market yields rise is making a discretionary choice. The company might argue that it is using a "settlement approach" or has a longer-duration obligation justifying a lower rate, but this requires explanation. If the footnote is silent, flag it.

The assumed return: when it is unrealistic

The assumed return on assets should be benchmarked to:

1. The company's own historical returns. Most companies disclose historical returns over 5, 10, or 20 years in the pension footnote. If the assumed return exceeds the 10-year historical return by more than 1%, question it.
2. Peer assumed returns. Many companies in the same industry will have disclosed their assumed returns in their filings. If the company's assumption is the highest in the group, investigate.
3. Market expectations. In a low-yield environment, assumes of 7%+ on diversified portfolios are hard to justify. In a higher-yield environment, 6.5% may be reasonable.

A company might argue that a higher assumption reflects a longer time horizon (pension assets are invested long-term) or a more aggressive asset allocation (more equities, fewer bonds). These arguments have merit but require disclosure. If the company simply assumes 7.5% without explanation, and peers assume 6.5%, red flag.

The deferred gain/loss tail wagging the dog

Pension accounting allows companies to defer the impact of actual returns differing from expected returns. Over time, actual returns diverge from assumptions, creating a "unrecognized gain" or "unrecognized loss" in accumulated other comprehensive income (AOCI).

A pension plan with a large unrecognized loss is a ticking time bomb. Eventually, this loss must be amortized into pension expense, hitting future earnings. A company with a large unrecognized loss and assumptions that are unrealistically optimistic is setting itself up for negative surprises.

To assess risk, check the footnote for:

• Unrecognized gains/(losses): A large negative number (loss) signals that actual returns have lagged assumptions by a wide margin. This is a ticking bomb.
• Average amortization period: Losses are typically amortized over 5–10 years. A company with a $500 million unrecognized loss and a 5-year amortization horizon will feel a $100 million hit to earnings (gross of tax) each year for 5 years.

Red flags: patterns to watch

Red flag 1: assumed return growing over time. If a company's assumed return has crept from 6% (five years ago) to 7.2% today, while peers remain flat and market expectations have declined, management is aggressively managing assumptions.

Red flag 2: assumed return consistently exceeds actual returns. The pension footnote discloses both the assumed and actual returns. If actual returns have averaged 5% over the past 10 years but the company assumes 7%, the assumption is disconnected from reality.

Red flag 3: discount rate not moving with bond yields. If 10-year corporate bond yields have risen 100 basis points in a year, the company's discount rate should rise roughly 50–100 basis points (depending on plan duration). If it hasn't, the company is cherry-picking the measurement convention.

Red flag 4: pension expense declining while plan size is stable. If pension expense has dropped 20% year-over-year while the obligation and headcount have not changed, assumptions have shifted. Investigate the change disclosures in the footnote.

Red flag 5: large unrecognized losses coupled with high assumed returns. A company with a $1 billion unrecognized loss (in AOCI) but an assumed return of 7.5% is being inconsistent. It is acknowledging massive past shortfalls while refusing to lower future assumptions. This is a tell.

How to read the pension footnote for assumption changes

The pension footnote always includes a table reconciling the opening balance of the pension obligation, contributions, payouts, and closing balance. A separate table shows the funded status (assets minus liabilities). Most importantly, a table discloses the key assumptions (discount rate, expected return on assets, wage growth, mortality).

Look for:

• Comparison of assumptions across years. Spot any changes (usually flagged in rows labeled "Change in assumed return" or similar).
• Discussion of changes. A well-written footnote explains why assumptions changed (e.g., "We lowered the discount rate to 4.1% from 4.3% to reflect current bond yields"). Vague or missing explanations are suspect.
• Unrecognized gains and losses. This section discloses deferred gains or losses, which should be considered when evaluating the reasonableness of current assumptions.

Example (simplified):

Assumption	2023	2022	Change
Discount rate	4.1%	4.3%	(20 bps)
Expected return on assets	7.0%	6.8%	+20 bps
Salary growth rate	3.0%	3.0%	–

In this example, the discount rate fell (raising liabilities and expense), but the expected return rose (lowering expense). The net effect on pension expense depends on the size of the plan, but offsetting changes like this warrant investigation. If market yields haven't moved, why did the company lower its discount rate while raising its expected return?

Real-world examples

Example 1: General Motors. GM has one of the largest pension liabilities in corporate America, with a multi-billion-dollar underfunded plan. Over the 2010s, as interest rates fell, GM was forced to lower its discount rate (raising the liability). To offset the income statement impact, GM gradually raised its assumed return on assets from 7.75% to 8.5% and later adjusted it based on market performance. However, actual returns have consistently lagged assumptions, creating a growing unrecognized loss. GM's strategy was to use assumptions to smooth earnings while the underlying pension position deteriorated.

Example 2: IBM. IBM has offered defined-benefit pensions to long-tenured employees for decades. As the company faced profitability pressure in the 2010s, it adjusted pension assumptions in ways that dampened the impact on earnings. Specifically, IBM extended the amortization period for deferred losses, slowing the recognition of pension losses on the income statement. While not technically an assumption change, the effect was similar: smoothing earnings by deferring the pension reality.

Example 3: AIG and the financial crisis. AIG's pension obligations ballooned during the 2008 financial crisis because it assumed that its pension assets would earn its expected return (e.g., 8%), but market losses meant actual returns were negative. AIG had to amortize massive unrecognized losses, hitting earnings. In hindsight, AIG's assumed returns had been unrealistically high for years; the crisis just forced recognition.

Common mistakes investors make

Mistake 1: assuming pension assumptions are objective. Many investors think pension accounting is rigid and technical, so they don't scrutinize it. In reality, assumptions are subjective judgments. A 1% change in the assumed return can swing annual earnings by $20–50 million for a large company.

Mistake 2: ignoring the pension footnote if the plan appears to be well-funded. Funded status can change quickly. A plan that is 90% funded today can be 75% funded in a down market. And even a well-funded plan can mask earnings smoothing through assumption changes.

Mistake 3: not comparing assumptions to peers. Assuming that a company's 7% expected return is reasonable without checking what peers assume is lazy. Peer comparison is the easiest red flag to spot.

Mistake 4: backing out pension items incorrectly when adjusting earnings. Some investors back out the entire pension expense line, assuming it is non-recurring. In reality, the service cost portion is recurring and legitimate. Only adjustments related to assumption changes should be backed out.

Mistake 5: ignoring discount rate changes tied to market movements. While discount rates are partly externally driven by bond yields, companies still have discretion in how they apply them (settlement approach, duration matching, etc.). A company can use legitimate reasoning to justify a lower rate, but the choice still suppresses reported expense.

FAQ

Q: Are pension assumption changes disclosed clearly in the financial statements?

A: They are disclosed in the pension footnote, but clarity varies. Some companies explain assumption changes well; others bury them. The key assumptions (discount rate, expected return on assets) are always shown in a table, but explanations of why they changed may be minimal. This is why forensic investors read the pension section carefully.

Q: What is a reasonable assumed return on pension assets?

A: A reasonable assumption depends on the portfolio mix and market conditions. A 60/40 stock/bond portfolio in today's environment might have expected returns of 5–6%. A 70/30 portfolio might justify 6–7%. The company's historical returns over 10+ years provide an anchor; an assumption more than 1% above historical returns is suspect.

Q: Can a change in pension assumptions be justified by a change in asset allocation?

A: Yes. If a company shifts its pension portfolio from 60/40 stocks/bonds to 70/30, a slightly higher expected return is justified. But the change should be disclosed and proportionate. A 100+ basis point increase in assumed return should require a meaningful portfolio shift or market change, not just a reallocation of 5–10 percentage points.

Q: How do I adjust reported earnings to account for discretionary pension assumption changes?

A: Identify the change in pension expense due to assumption changes (disclosed in the footnote or calculated by comparing the pension expense component of the current year to the prior year, holding asset value and obligation size constant). Back out the after-tax benefit or charge from reported net income. This gives adjusted earnings.

Q: What is the relationship between pension expense and the funded status?

A: Pension expense on the income statement is one thing; the funded status (assets minus liabilities) on the balance sheet is another. A company can reduce reported expense through assumptions but still have a deteriorating funded status. The two should be cross-checked. A company with declining funded status but declining reported expense is likely using assumptions to mask deterioration.

Q: Are pension adjustments commonly used by companies to beat earnings targets?

A: Yes, though it is difficult to prove intent. Academics have documented correlations between pension assumption changes and earnings pressure, and between assumption changes and executive compensation tied to earnings targets. Pension assumptions are one of the most common and least-scrutinized levers for earnings management.

Related concepts

• Defined-benefit vs. defined-contribution plans: The shift from DB to DC plans has reduced pension exposure for many companies, but large legacy plans still carry significant assumptions.
• Other post-employment benefits (OPEB): Companies also offer retiree healthcare, which is accounted for similarly to pensions and subject to the same discretionary assumptions.
• Accumulated other comprehensive income (AOCI): The balance sheet account where unrecognized pension gains and losses live; over time, these must reverse and hit earnings.
• Mortality tables and salary growth assumptions: These demographic assumptions are less frequently manipulated than return assumptions but can still be tweaked to manage reported earnings.
• Pension settlement and curtailment: Companies sometimes buy annuities to settle pension obligations, eliminating the liability and creating a one-time accounting gain. This is legally different but serves a similar earnings-management purpose.

Summary

Pension accounting is a minefield of discretionary judgment. The discount rate and assumed return on assets are two of the most powerful levers for managing reported earnings, and they are among the least scrutinized by investors. A company that raises its assumed return, lowers its discount rate, or offsets these changes while operational performance is declining is likely using assumptions to smooth earnings.

To spot red flags:

1. Compare the company's pension assumptions to prior years and to peers. Outlier assumptions warrant investigation.
2. Benchmark the assumed return to the company's historical returns. If the assumption exceeds the 10-year average by more than 1%, question it.
3. Monitor the unrecognized gain/(loss) balance in AOCI. A growing loss signals that assumptions are becoming increasingly disconnected from reality.
4. Read the pension MD&A discussion carefully. Vague explanations of assumption changes are more suspicious than detailed ones.
5. Back out the impact of assumption changes when trending earnings across years or comparing to peers.

Pension assumptions are where some of the most sophisticated earnings management happens. Investors who ignore them are missing a critical red flag.

Next

Supplier-financing arrangements as a red flag →