Actuarial Neutrality

Definition

Actuarial neutrality is a property of a pension system in which the implicit tax or subsidy on continued work is zero at every age within the statutory retirement window. A system is actuarially neutral if an individual's expected lifetime benefit — discounted at the actuarial rate — rises by exactly the present value of the forgone pension during an additional year of work, so that delaying retirement by one year leaves the individual's total wealth unchanged. This is distinct from actuarial fairness (Net Present Value Ratio, NPVR = 1), which requires lifetime benefits to equal the actuarial value of lifetime contributions. A system can be actuarially fair in the aggregate but impose large marginal taxes or subsidies on the retirement timing decision at particular ages.

The implicit tax/subsidy rate at retirement age $x$ is:

$\text{TAX}(x) = \frac{\text{SSW}(x) - \text{SSW}(x+1)}{w(x+1)} - 1$

where $\text{SSW}(x)$ is Social Security Wealth (the present value of expected future benefits if retiring at $x$ ), $\text{SSW}(x+1)$ is SSW if retiring one year later, and $w(x+1)$ is the foregone net wage during the additional year. $\text{TAX} = 0$ means the increment in SSW from working an additional year exactly offsets the contribution made and the benefit delayed.

Key Ideas

Actuarial neutrality vs. actuarial fairness: Actuarial neutrality is a marginal concept (no incentive at any single retirement age); actuarial fairness is a lifetime concept (NPVR = 1 on average). Both are needed for a fully distortion-free system: fairness removes intergenerational redistribution, neutrality removes inframarginal retirement age incentives.
Italy DB: large distortions: Under the pre-1995 Italian Defined Benefit (DB) system, TAX reached 50% at some ages — an implicit 50-cent tax for every dollar of net wage earned in an additional year of work — creating enormous incentives for early retirement and low labor force participation among older workers (Belloni and Maccheroni 2006).
Italy NDC: approximate neutrality: The 1995 Dini reform replaced DB with a Notional Defined Contribution (NDC) system. By design, NDC links benefits to accumulated contributions, so TAX falls to approximately 4% — close to, but not exactly, zero.
Period table distortion: NDC transformation coefficients are computed from period life tables (Italy: 1990 Istituto Nazionale di Statistica (ISTAT) tables, updated every 10 years). Under improving mortality, cohort life expectancy exceeds period life expectancy, so the expected annuity is underpredicted → transformation coefficient is set too high → pension is systematically overpaid → NPVR > 1. This overpayment was estimated at ≈ 6 pp of NPVR for Italy; roughly two-thirds attributable to the period/cohort divergence, one-third to the 10-year revision lag. See Period vs. Cohort Life Expectancy.
Pre-revision spikes: In years immediately before a scheduled 10-year revision of transformation coefficients, workers face a TAX spike of 30–40 pp: delaying retirement until after the revision yields a higher coefficient (reflecting improved mortality). This discontinuity in incentives is a direct artefact of infrequent discrete coefficient updates.
Intragenerational redistribution shift: DB systems redistribute toward higher earners (longer contribution histories, higher wages). NDC systems neutralize this but introduce a new redistribution dimension: toward longer-lived individuals. Since women outlive men, equal contributions yield higher SSW for women under NDC — a transfer from men to women absent under DB.

How It Works

In a DB system, the benefit formula is set by rule (e.g., accrual rate × average wage × years of service) and is not automatically sensitive to the actuarial cost of delaying retirement. If the formula is generous at early ages and benefit growth slows at older ages, TAX becomes large and positive, creating an implicit retirement tax.

In an NDC system, the pension equals the accumulated notional fund multiplied by a transformation coefficient:

$P = F \cdot \delta_x$

where $F$ is the notional fund (contributions + notional returns) and $\delta_x = 1/\mathbb{E}[a_x]$ is the inverse of the expected annuity value at age $x$ (incorporating survival probabilities and a discount rate). By construction, delaying retirement by one year increases $F$ by an additional year's contribution and notional return, while $\delta_{x+1}$ adjusts for the shorter remaining life. Under correct actuarial tables, these effects cancel and $\text{TAX} \approx 0$ .

The failure of neutrality arises when $\delta_x$ is miscalibrated — specifically, when period life tables are used instead of cohort (prospective) tables. Period tables understate remaining life expectancy under improving mortality, so $\delta_x$ is set too high, delivering more pension per unit of fund than is actuarially warranted.

Why It Matters

Retirement age distortion: Non-neutral systems push workers to retire earlier (or later) than is efficient, reducing labor supply at the extensive margin for older workers. Italy's pre-reform TAX of up to 50% is a plausible contributor to Italy's historically low employment rates for workers aged 55–64.
Fiscal sustainability: NPVR > 1 (actuarially unfair NDC) means the system pays out more than it collects in present value terms, eventually requiring either contribution rate increases, benefit cuts, or deficit financing.
Incentive design under longevity growth: The 10-year revision lag and period-table bias mean that as longevity improves, NDC systems automatically become more actuarially unfair. Continuous updating or the use of cohort (prospective) mortality tables would eliminate this problem.
Comparison with DB reform debate: The NDC architecture is often presented as a "neutral" solution that avoids the incentive problems of DB. The Belloni-Maccheroni analysis shows this claim is only approximately correct and depends critically on the quality and currency of the mortality tables used.

Open Questions

Would a continuous updating mechanism (annual, not decennial) eliminate the pre-revision spikes? The answer is yes mechanically, but raises political economy questions about benefit predictability.
How large is the period-cohort mortality gap for other NDC countries (Sweden, Poland, Latvia)? Italy's gap may differ from countries with different improvement rates.
Is the intragenerational redistribution shift (toward longer-lived) normatively desirable? It could be framed as fair (you get more because you live longer) or unfair (men subsidize women's longer lives).
Does actuarial neutrality actually affect retirement timing? The empirical labor supply response to TAX in Italian data is not directly estimated in this paper; the behavioral implication is inferred from theory.

Sources

Belloni and Maccheroni 2006 — Actuarial Fairness when Longevity Increases