This paper proposes a new way to design Target-Date Funds (TDFs) so they aim for a clear retirement return while controlling risk directly. Instead of setting age-based limits on stocks and bonds, the regulator or sponsor specifies a declining Conditional Value-at-Risk (CVaR) glidepath. CVaR is a portfolio-level measure of how bad losses can be on the worst days. The glidepath tells managers how much tail risk the fund may take at each point in time while trying to reach a chosen return target.

The researchers make the target return an input set outside the model. It can come from retirement choices such as retirement age, contribution rate, years worked, life expectancy, and the desired replacement rate. They then calibrate a time series of CVaR limits so that the risk taken at each age is consistent with trying to meet that return with high probability. A key modelling choice is conservative: each month the manager is assumed to pick an allocation at random from the set of portfolios that meet the CVaR limit. This gives success probabilities that are averages over all allowed allocations rather than best-case outcomes.

To compare candidate glidepaths the authors introduce two simple metrics. One is the probability of meeting the target return. The other is the cumulative risk taken over the fund’s life. As a proof of concept they apply the framework to Chile’s 2025 pension reform. They use nine Chilean and global asset classes and a 40-year accumulation horizon. The Chile case is relevant because the reform will make enrollment mandatory and assign workers to generational funds (fondos generacionales) labeled A to E, with only the choice of asset manager left to workers.

Two findings stand out in the Chile example. The age at which the fund begins to reduce risk — the transition age — is the single most important design choice. Glidepaths that de-risk too early tend to fail to deliver the required return, no matter how risky they are earlier on. The second finding is that contribution density acts as a hard limit: if contribution rates are below a critical level, no glidepath can make up for them and the target becomes unreachable.