Ronald Lee and Shripad Tuljapurkar (1997)
“Death and Taxes: Longer Life, Consumption and Social Security”

This is a somewhat technical article.  I will try to summarize the main ideas in plain English, at the risk of oversimplifying the argument.

Main idea: The age structure of the US population is changing, with more retirees relative to the numbers of children and workers.  This has implications for the funding of Social Security.  In the past, demographers have focused upon fertility change as an explanation for the changing age structure.  The authors want to study the effect of mortality decline (increasing life span).

In a stable population, mortality affects the age distribution in two ways.

In a population where the life span is low, an increase in the life span makes the population younger because more people survive to bear children.  In a population like the modern US, nearly everyone survives to reproductive age, so gains in life span only make the population older.

Three stages of the “economic life cycle”:

Mortality decline affects the percentage of life spent at each stage of the cycle.  The authors assume that every individual faces a life cycle budget constraint (a sum total of lifetime earnings, which the individual allocates to spend at different times over the course of his or her entire life).  Mortality decline affects the constraint in two ways.  Some of the additional years of life (in a population) are added to the working years, which loosens the budget constraint (more years at work equals higher lifetime income).  Other years are added to retirement, tending to tighten the budget constraint.
Based on the actual US age distribution in 1995, the authors calculate that a 1 year increase in average lifespan (from 75.5 to 76.5 years) would add an extra year to the retirement phase of most people’s lives (see figure 2, page 70).  In other words, it tightens the budget constraint.  On average, people would need 0.8% more wealth to maintain their consumption levels.
The authors discuss a variety of ways in which this calculation is not completely realistic…
They also discuss the work of economists who came to similar conclusions based on different sets of assumptions…

The same calculation can assess the impact of increased life spans on Social Security.  Since most Americans already survive to retirement, an extra year of life increases benefits far more than taxes (see figure 3, p. 73).  Based on 1994 data, a one-year increase would require a tax hike (or benefit reduction) of 3.6%.

The Social Security Administration (SSA) expects US life expectancy to be 80.7 years in 2070.  The authors consider this an extremely low forecast.  SSA is assuming mortality will fall much more slowly in the 21st century than in the 20th.  But international comparisons show that countries with longer life spans than the US (such as Japan) have recently achieved rapid gains.  So there is no reason to believe the US rate of mortality decline will slow down soon.

Using different estimates of the rate of mortality decline between 2000 and 2070, the authors calculate the tax rate that will be required to balance Social Security’s budget.  They find that while the balanced budget tax rate is guaranteed to go up (given that everyone forecasts increasing life spans in the US), its 2070 value is very sensitive to different estimates of mortality decline (see table 2, p. 77).

The authors discuss why their method of forecasting mortality is better than SSA’s method…