Fat-tail stochastic model

What “fat tail” means

A fat-tailed return distribution allows very large positive and negative shocks to occur more often than they do in the standard bell-shaped, or normal, distribution. The name refers to the ends—or tails—of the distribution, not to the average result. A model can have the same expected return and volatility inputs as another model while placing more probability in extreme outcomes.

In retirement planning, the negative tail deserves particular attention because a deep loss early in retirement can force withdrawals from a reduced portfolio. The same sequence can leave less capital available for a later recovery. The fat-tail model is therefore a way to ask, “What if unusually severe market moves occur more often than the standard model assumes?”

When a fat-tail comparison is worth running

Run this comparison when the plan is close to its success threshold, depends heavily on market growth, begins withdrawals soon, or appears vulnerable to early losses. It is also useful when two strategies look similar under standard Monte Carlo and you want to see whether one is more resilient to extreme shocks.

A worse fat-tail result does not automatically mean the plan or allocation must change. It tells you that the conclusion is sensitive to the assumed shape of returns. The appropriate response may be to build more margin, keep a larger flexible-spending option, delay a commitment, or simply recognize the uncertainty before deciding.

Set up a fair standard-versus-fat-tail comparison

Start with a saved plan whose balances, spending, retirement dates, income, allocation, and planning horizon are current. Use enough trials for the reported percentiles to be reasonably stable. Keep the same run count, seed, expected stock, bond and cash returns, volatility, correlations, inflation assumptions, guardrails, and lifespan settings in both runs.

Change only Return shock distribution from Standard model to Fat-tail stress for the first comparison. If several controls change at once, you will not know whether a different result came from the distribution or from an unrelated assumption.

Run the fat-tail stress in YARCalc

Open Stochastic Analysis and run the standard Monte Carlo model first. Record the success probability, P10 and median ending balances, worst drawdown, first depletion years, and any spending cuts or shortfalls that matter to the household.

Open the analysis settings, keep Monte Carlo selected, and choose Fat-tail stress under Return Distribution. The Tail degrees of freedom setting controls how heavy the tails are. Lower values create more frequent extreme shocks; higher values move the distribution closer to the standard model. The allowed range is 3 through 30, and the default is 5.

Run the stress with the same saved plan and comparison settings. Put the two results side by side. The most useful question is not whether every number got worse, but where the plan’s behavior changed enough to affect a decision.

What changes inside the simulation

YARCalc starts with correlated stock, bond, cash, and inflation shocks. Under Fat-tail stress, a common Student-t scale factor occasionally makes all of those correlated shocks unusually large. The scale is adjusted so that the variance remains comparable to the selected volatility assumptions. The stress changes the frequency of extremes rather than simply raising every year’s volatility.

Because the same scale affects the correlated shocks in a year, a tail event can influence several parts of the plan together. Asset returns are still bounded by the application’s return limits, inflation cannot fall below the model’s floor, and all of the normal tax, withdrawal, spending-response, healthcare-shock, and major-expense rules still run.

Fat tails create both extreme upside and extreme downside. Retirement results often react more strongly to the downside because withdrawals after a loss can make recovery harder. That asymmetry can lower success probability or downside balances even when the model also creates unusually strong positive years.

Read the comparison beyond one success percentage

First compare downside measures: P10 ending balance, severe drawdown, depletion timing, shortfall frequency, and the amount of spending reduction. Then compare the median and upper percentiles to see whether the entire distribution moved or mainly became wider.

Look at when failures occur. A concentration of early failures points toward sequence risk and may make cash reserves, near-term spending flexibility, retirement timing, or allocation transitions relevant. Failures only near the end of a very long horizon may point instead toward longevity, inflation, or the size of the legacy goal.

Repeat the comparison with the same seed if you are checking a setup change. If a conclusion depends on a very small difference, increase the run count or repeat with other seeds before treating it as durable.

What the fat-tail stress does not tell you

The model does not predict when a crash, inflation shock, recovery, or unusually strong market will happen. Degrees of freedom is a statistical severity control, not a forecast based on current news and not a statement that a named historical crisis will repeat.

A fat-tailed statistical distribution still simplifies markets. It does not reproduce every liquidity freeze, tax-law change, trading constraint, correlation breakdown, prolonged regime, or personal response that can accompany a real crisis. Historical bootstrap, Markov regimes, forced bad-sequence settings, and explicit spending What-ifs answer different stress questions.

Treat the fat-tail run as one robustness check. A plan that survives it is not guaranteed, and a plan that struggles under it is not certain to fail. The value is seeing which decisions remain reasonable under a more demanding description of uncertainty.

Continue on the screen that uses this concept

Use this reference with the Stochastic Analysis workflow rather than as a stand-alone plan input.