What the evidence says
The evidence base for this topic is demographic and institutional, not US-state-specific: every source here covers a national pension system (Central America regionally, an EU-wide review, and Spain in detail) or a general simulation model. Across all three, population aging is the common driver of financing pressure. Central America's working-age-to-elderly population ratio decreases from about 8:1 today to less than 3:1 by 2050. In the EU, life expectancy at birth increases by 5.3 years for men and 5.1 years for women between 2016 and 2070, and declining fertility combined with rising longevity increases the age dependency ratio. In Spain, this demographic shift translates directly into fiscal pressure: pension expenditure as a share of GDP increases markedly in the second half of the century with no parallel increase in system revenue, producing what the source calls acute financing shortfalls. None of these sources quantify the exact size of the fertility-longevity-to-dependency-ratio relationship; the direction is consistent but the magnitude is not established outside Central America and the EU life-expectancy figures.
On benefit design, one regional source finds that Central American defined-benefit systems pay a real internal rate of return of about 10% on average, more than twice the real growth rate of the wage bill that funds them — a gap the source attributes to benefits that are generous relative to contributions and retirement ages that are low relative to international comparators, though the size of that international gap is not quantified.
On institutional responses, the evidence describes several design patterns without measuring their effects: European countries (including France, Greece, Hungary, Romania, and Spain) adopted parametric reforms — benefit reductions and retirement-age increases; Italy, Latvia, Poland, and Sweden introduced Notional Defined Contribution schemes starting in the 1990s; and several OECD countries (Australia, Canada, Norway, Sweden, Latvia, Poland) combine funded and pay-as-you-go elements within mandatory systems. Spain specifically has run a dedicated trust fund since 2000 that holds pension-system surpluses in fixed-income instruments — the closest example in these sources to a "dedicated funding stream." None of the sources measure how much these designs reduce unfunded liabilities.
One source does measure a mechanism with direct bearing on dedicated funding streams: Spain's 2013 automatic adjustment mechanism, which indexes pensions to changes in life expectancy and to the system's financial condition rather than to price inflation, decreases the system's remaining financial imbalances substantially compared to a system without it. But the same mechanism acts only once flow imbalances appear and ignores the system's accumulated implicit liabilities, and nearly all the resulting gains are captured by future labor-market entrants while nearly all the burden falls on current workers and, to a lesser extent, current retirees.
A separate simulation-based line of evidence examines the risk of relying on invested surplus to cover a required contribution increase: under model parameters of 4% drift and 20% volatility, a one-year investment achieves only about a 58% probability of fully repaying the state. Reaching 99% probability at that horizon requires investing 53% more than the amount needed to cover the contribution increase directly. This risk falls as the investment horizon lengthens. These figures come from a calibrated stochastic model, not observed fund performance.