Notional Defined Contribution

Definition

A Notional Defined Contribution (NDC) system is a pay-as-you-go (PAYGO) public pension scheme structured to mimic the mechanics of a funded defined-contribution (DC) plan. Each worker has an individual "notional" account to which contributions are credited, earning a notional rate of return equal to the growth rate of gross domestic product (GDP) (or the covered wage bill) rather than actual financial market returns. At retirement, the accumulated notional fund is converted to an annual pension using a transformation coefficient ( $\delta_x$ ) that reflects the expected actuarial value of remaining life at the chosen retirement age. Unlike a funded DC plan, no actual assets back the notional accounts; redistribution between cohorts happens through the PAYGO mechanism.

Italy adopted NDC in the 1995 Dini reform; Sweden adopted a closely related system in 1994. Poland, Latvia, and several other countries have since followed.

Key Ideas

Notional fund accumulation: Each year, a worker contributes a fraction of wages to their notional account. The account earns a return equal to the growth rate of the covered wage bill (Italy) or GDP. At retirement age $x$ , the accumulated fund $F_x$ is the sum of all contributions compounded at these notional returns.
Transformation coefficient ( $\delta_x$ ): The pension is $P = F_x \times \delta_x$ . The coefficient $\delta_x = 1/a_x$ , where $a_x$ is the expected present value of a one-unit life annuity beginning at age $x$ . In Italy, $a_x$ is computed from the 1990 Istituto Nazionale di Statistica (ISTAT) period life tables and updated every 10 years by legislative decree. A higher $\delta_x$ $\to$ higher pension per unit of fund.
Actuarial fairness by design: Because benefits are explicitly tied to accumulated contributions and to expected remaining life at retirement, NDC systems achieve net present value ratio (NPVR) $\approx 1$ almost automatically — unlike defined benefit (DB) systems whose benefit formulas are not actuarially calibrated. See Actuarial Neutrality.
Actuarial neutrality approximately achieved: The $\text{TAX}$ on continued work under NDC is approximately zero: working an additional year increases the fund by contributions while $\delta_{x+1}$ adjusts for the shorter annuity. In Italy post-reform, $\text{TAX} \approx 4\%$ (compared to up to $50\%$ under the old DB) (Belloni and Maccheroni 2006).
Period-table bias: The critical weakness of Italy's NDC implementation is that transformation coefficients rely on period rather than cohort mortality tables. Under improving mortality, cohort life expectancy (LE) $>$ period LE, so $a_x$ is underestimated, $\delta_x$ is set too high, and the actual pension exceeds the actuarially fair level. This generates a persistent $\approx 6$ percentage point (pp) NPVR overpayment — approximately two-thirds from the period/cohort divergence and one-third from the 10-year revision lag. See Period vs. Cohort Life Expectancy.
Pre-revision spikes: Because $\delta_x$ is updated discretely every 10 years, workers near a revision date have an incentive to delay retirement until after the update (which will set a new, typically lower $\delta_x$ reflecting improved longevity). This creates $\text{TAX}$ spikes of $30$ – $40$ pp immediately before each revision.
Long transition in Italy: The Dini reform applied NDC fully only to workers with zero pre-1996 contributions. Workers with mixed careers receive a pro-rata blend of DB and NDC. Full NDC coverage reaches the majority only around 2030.
Intragenerational redistribution: NDC shifts redistribution from a wage-based pattern (DB favors high earners) to a mortality-based pattern (NDC favors longer-lived individuals). Since women outlive men, they receive higher social security wealth (SSW) per euro of contribution under NDC.

How It Works

The Italian NDC formula: at retirement age $x$ , the pension $P = F_x \times \delta_x$ , where:

$F_x = \sum_t c_t \cdot w_t \cdot \prod_{s=t}^{x-1}(1+g_s)$ — contributions $c_t \times$ wage $w_t$ , compounded at wage bill growth $g_s$
$\delta_x = 1/a_x$ , where $a_x = \sum_{k=0}^{\omega-x}(1+r)^{-(k+1)} \cdot S(x,\, x+k)$ — expected present value of annuity, discounted at rate $r$ and weighted by period survival probabilities $S(x, x+k)$ from the ISTAT tables

The actuarial neutrality condition $\text{TAX}(x) = 0$ holds when $\delta_x$ correctly reflects the actual future mortality of the retiring cohort. If period tables understate $S(x, x+k)$ (because future mortality will be lower than current period rates), then $a_x$ is underestimated, $\delta_x$ is overestimated, and $P$ is larger than warranted — overpayment persists until the next table update.

Why It Matters

Pension reform template: NDC has been widely adopted or proposed as a replacement for unaffordable DB systems because it aligns benefits with contributions and removes distorted retirement incentives. Italy's experience shows the design details (choice of mortality tables, update frequency) matter enormously for whether the reform actually achieves its stated goals.
Longevity risk allocation: Under NDC with fixed transformation coefficients (as in Italy), longevity risk is borne by the government / future taxpayers — because if people live longer than projected, the system overpays. Switching to prospective (cohort) mortality tables, or to continuous updating, would shift longevity risk back to retirees (lower coefficients → lower pensions as longevity improves).
Transition costs and DB legacy: Italy's long transition means the old DB system's actuarial unfairness persists for the majority of current retirees. The NDC benefits will materialize gradually, not immediately.
Comparative pension design: NDC is sometimes confused with funded DC (different: NDC has no actual assets) and with "notional" accounts in general (NDC is a specific PAYGO design, not just any account structure). The Sweden 1994 and Italy 1995 implementations differ in key parameters (notional return index, annuity rate used).

Open Questions

Should transformation coefficients use prospective cohort tables rather than period tables? Cohort tables better reflect actual longevity but require mortality projections with inherent uncertainty (see Period vs. Cohort Life Expectancy). Who should bear that projection risk?
Should the NDC notional return equal GDP growth, wage bill growth, or the system dependency ratio (as proposed in some Swedish-style reforms)? The choice affects intergenerational fairness.
Does the NDC structure in practice change retirement behavior in Italy? The empirical retirement timing response to the $\text{DB} \to \text{NDC}$ incentive change is separate from the theoretical actuarial analysis in Belloni and Maccheroni.

Sources

Belloni and Maccheroni 2006 — Actuarial Fairness when Longevity Increases