Age-Rank

Why age-based rank within organizations persists

Motivation

Within any age cohort, the variance in individual skill, judgment, and drive is large—often larger than the difference between age cohorts separated by a decade of experience. Therefore, one might expect a competence-driven organization to produce a hierarchy that is substantially age-mixed. But this is rarely the case. In practice, most organizations are ordered by age far more strongly than would be predicted based on variance in skill. Something other than competence appears to be doing much of the work.

Several mechanisms are at play here, including but not limited to the following.

Before enumerating them, it is worth flagging a more general pattern. Social systems of any complexity tend to develop a dominant ordering along some scalar attribute; dominance hierarchies are near-universal among social species, and informal status rankings re-emerge even in organizations that aim to be flat. Some total ordering is likely to form whether or not everyone intends it. Age is an especially natural attribute around which such an ordering settles, because it is observable, is monotonically increasing, and does not require adjudicating everyone’s competence. The three mechanisms below explain why, once the ordering has fallen on age, it is difficult to dislodge.

In-group solidarity. Once a senior cohort exists, its members share an interest in preserving the current ordering, by vouching for one another, coordinating informally around who rises, etc. The ordering is self-reinforcing because the people with influence over the next round of promotions benefited from the previous round’s outcomes. Simply put, humans also feel a visceral comfort and kinship with those closer in age; they just relate. (This mechanism is formalized in Scene 3 below.)

Payscale insurance. A payscale that rises with age is itself a structural insurance arrangement. Younger employees support it because it promises future upside; older employees support it because they are currently drawing on it and will go on to earn the highest salary when they are latest in their careers—exactly the time when personal competitiveness and competence are mostly liable to decline. This means that if older workers manage to hang on even modestly past what a competence-justified tenure would have been, they reap outsized returns. The result is a shared incentive to preserve the age-based ordering that does not require explicit coordination.

Pulling up the ladder (holdup). Older employees accumulate institutional knowledge, relationships, and procedural access whose value depends on the incumbent’s cooperation. A senior who is bypassed by a rookie (or under threat of such) can reverse or arrest that rookie’s further ascent without direct sabotage (e.g., by withholding context, declining to make introductions, or being slow to respond). The threat does not need to be exercised to shape decisions; its availability is enough to make organizations reluctant to promote against the age ordering in the first place.

Together, these non-mutually-exclusive forces can sustain what looks like a strong equilibrium around the age-based ordering. That equilibrium is common and well-documented; other equilibria remain possible, and different organizations settle at different points along the competence–age continuum. The goal of this simulation is to represent each force as a separate term with its own parameter, so that the conditions under which the age-based equilibrium dominates, weakens, or gives way can be examined directly.

The Core Idea

Each person in the simulation has a fixed age (read loosely as years of relevant experience) and a fixed competence (how good they are). Their rank evolves over time, pulled by three competing forces plus random noise.

A note on age vs. tenure. These are different quantities: a 45-year-old hired from a competitor firm has high age but zero tenure at the new employer. The purely firm-specific pressures (institutional leverage, the ladder-pulling mechanism) are carried by tenure; the other two mechanisms operate along both age and tenure, with a real sociopsychological component to the age dimension in particular, since people feel viscerally that something is strange or awkward when they observe stark exceptions to age-based ranking. For clarity, the simulation collapses the two into a single time variable and calls it age, which readers can interpret loosely as biological age, years in the field, or time at the firm. A fuller model would split the two and weight them separately, without changing the qualitative story.

In most real organizations, the age-ordering force dominates. But the simulation also lets you dial up each force independently to see why and to find the regimes where competence-based ordering can be sustained.

The Equation

$$dr_i = \bigl[\,\alpha(c_i - \bar{c}) \;-\; \beta(r_i - s_i) \;-\; \gamma H_i\,\bigr]\,dt \;+\; \sigma\,dW_i$$
$$H_i = \tfrac{1}{N}\sum_{j}\max(0,\,a_j - a_i)\cdot\max(0,\,r_i - r_j)$$

Read it as: rank change each instant = ±competence pull ± age spring − holdup drag ± luck.

A subtlety on the $\alpha$ vs. $\beta$ comparison: the competence signal $(c_i - \bar{c})$ has standard deviation on the order of $0.18$, while the age target $s_i$ is spread across $[-1, 1]$ with standard deviation closer to $0.58$. The two forces act on signals of quite different scales, so even at $\alpha = \beta$, age exerts the stronger pull in practice. The $\alpha/\beta$ ratio should be read with that asymmetry in mind.

Payscale insurance maps to $\beta$; ladder-pulling to $\gamma$; in-group solidarity to the network lift $\lambda$ in Scene 3. The $\alpha$ term stands apart—it is the pull of competence, the one force working against the age-based equilibrium rather than sustaining it.

Every Term Explained

$\alpha(c_i - \bar{c})$
Competence drift α slider
Pulls your rank up if you're above-average competence, down if below. α controls how much the org actually rewards ability. In most orgs, α is small: talent is noticed slowly and incompletely.
$-\beta(r_i - s_i)$
Age restoring force β slider
Pulls your rank back toward $s_i$, the rank your age alone predicts. Think of it as a spring: the further you've drifted from "your place in line," the harder it pulls you back.

What three forces produce this “spring”? (1) Deferred compensation: firms underpay juniors and overpay seniors on purpose; the future overpayment is a bond that breaks if the age order is skipped Lazear 1979. (2) Tournament prizes: the pay gap between ranks is the incentive for everyone below; inverting rank compresses the prize Lazear & Rosen 1981. (3) Fairness norms: everyone watching recalibrates their expectations when the line is jumped Akerlof & Yellen 1990.
$-\gamma H_i$
Ladder-pulling γ slider
The formal version of the ladder-pulling mechanism described above. $H_i$ is nonzero only when junior $i$ is currently outranking some senior $j$—meaning a displaced senior can push back. The force on $i$ is proportional to how far above $j$ they’ve risen and how much more senior $j$ is.

The tournament literature shows that large rank-prize spreads create incentives to impede rivals rather than outperform them Lazear 1989—a senior threatened by a rising junior faces this incentive directly. Withholding cooperation, context, and access makes the junior’s contributions less legible to the org without any overt act required. Economists label the structural leverage this creates insider holdup Lindbeck & Snower 1988—though their original framing is the labor market, the within-firm mechanism is the same.
$\sigma\,dW_i$
Noise σ slider
Random shocks: politics, visibility, project luck, manager turnover. Not essential to the dynamics; set $\sigma \approx 0$ for clean motion.

The Variables on Each Dot

VarNameWhat it means
$a_i$AgeFixed. Determines x-position on the scatter. Read as years of relevant experience; see the note above on age vs. tenure.
$c_i$CompetenceFixed. Shown as dot color (blue = low, red = high). Drawn from a distribution at start.
$r_i$RankThe state variable; the only thing that moves. Determines y-position. This is what the SDE governs.
$s_i$Age targetThe rank $a_i$ predicts based on age order alone. The spring pulls $r_i$ toward $s_i$.
$H_i$Holdup forceComputed from the current positions of all other agents. High when junior $i$ is sitting above many seniors.

Parameter Regimes

Age dominates
$\alpha/\beta = 0.2$,  $\gamma$ moderate
After mutiny, cloud snaps back to diagonal fast. Blue $\tau(\text{rank}, \text{age})$ recovers fully. The normal org.
Competence wins
$\alpha/\beta = 6$,  $\gamma \approx 0$
Dots sort by color (red up, blue down). Orange $\tau(\text{rank}, \text{comp})$ stays high. Age order is irrelevant.
Ladder-pulling
$\gamma = 2.0$,  $\beta = 0.2$
β is weak, so recovery after mutiny is driven by holdup alone: slower and lumpier, but it still wins.
Phase boundary
$\alpha/\beta \approx 2$,  $\gamma$ small
The actual phase boundary, accounting for the scale asymmetry. Neither attractor dominates: after mutiny, both $\tau$ values fluctuate and the outcome is sensitive to noise.

Scene 2: The Adaptive Layer

Each senior now carries a hoarding intensity $h_i \in [0,1]$, measuring how hard they're working to withhold institutional knowledge. The gold ring around a senior dot shows $h_i$.

$$\frac{dh_i}{dt} = \varepsilon \cdot v_i \;-\; \delta \cdot h_i$$
$v_i$ = number of juniors currently outranking senior $i$

When a mutiny strikes, violation counts spike → hoarding intensifies → holdup force strengthens → age order is restored faster → violations drop → hoarding decays. The equilibrium defends itself endogenously. This is why a one-off competence-based reshuffle of a company can easily fail; it triggers incumbents to employ self-protective behaviors that reverse those improvements.

Scene 3: The Solidarity Network

Each agent is embedded in a social graph whose edges represent relationships—shared cohort, shared history, informal alliance. A neighbor with higher perceived standing exerts upward lift; the force is asymmetric, since those already surpassed offer nothing further. This is the formal rendering of in-group solidarity.

$$\text{lift}_i = \frac{\lambda}{N-1}\sum_{j \,\sim\, i} \max(0,\, p_j - p_i)$$
$p_k = (1-\rho)\,r_k + \rho\,c_k$  ·  $j \sim i$ = neighbor of $i$  ·  $\lambda$ = lift strength  ·  $\rho$ = recognition weight (default 0)

Three graph topologies are available. Age-banded graphs pack edges within cohorts—the homophily case, where people cluster with those close in age. Small-world graphs add sparse cross-cohort bridges, capturing the mentor or sponsor relationship that ties a junior to a distant senior. Random graphs serve as the null: uniform connectivity, no age structure. The age-banded topology reinforces the age-based equilibrium; cross-cohort relationships are the mechanism by which a junior employee possibly gains standing beyond what their age cohort alone would supply.

The Key Ratio to Remember

$\alpha/\beta$ is the most useful single summary. Below ~0.5: age-based ordering wins. Above ~2.0: competence wins. In between: mixed, sensitive to perturbations. Most real organizations sit around 0.1–0.3. However, it very well may happen that $\alpha/\beta > 2$ at some companies. This might correspond to organizations that have structurally committed to performance feedback, and the model explores how that ordering can be self-sustained.

References

  1. Lazear, E.P. (1979). Why is there mandatory retirement? Journal of Political Economy, 87(6), 1261–1284.
  2. Lazear, E.P. & Rosen, S. (1981). Rank-order tournaments as optimum labor contracts. Journal of Political Economy, 89(5), 841–864.
  3. Lazear, E.P. (1989). Pay equality and industrial politics. Journal of Political Economy, 97(3), 561–580.
  4. Lindbeck, A. & Snower, D.J. (1988). The Insider-Outsider Theory of Employment and Unemployment. MIT Press. (See also: Lindbeck & Snower, Journal of Economic Perspectives, 15(1), 165–188, 2001.)
  5. Akerlof, G.A. & Yellen, J.L. (1990). The fair wage-effort hypothesis and unemployment. Quarterly Journal of Economics, 105(2), 255–283.
  6. Pluchino, A., Rapisarda, A. & Garofalo, C. (2010). The Peter Principle revisited: A computational study. Physica A, 389(3), 467–472.
  7. Kuran, T. (1989). Sparks and prairie fires: A theory of unanticipated political revolution. Public Choice, 61(1), 41–74.