Social Security and the coming Productivity Explosion
submitted as an op-ed to the NY Times
7 Nov 2004The debate over social security is based on projections as far as 75 years into the future. The Social Security Trustees estimate that withdrawals will exceed contributions by the year 2018, that the trust fund will be depleted by the year 2042, and that the total shortfall over 75 years will be 3.7 trillion dollars. These alarming estimates are being used to justify a radical change to social security using individual investment accounts. But because of their long time range, these estimated shortfalls become surpluses with small increases in the assumed rate of future productivity growth in the U. S. economy. The Trustees assume that productivity will grow at 1.6% per year from 2012 until 2078, whereas productivity has been growing at 2.5% over the past ten years. And rather than slowing down, the rate of productivity growth is likely to accelerate.
In fact, some time during the next 75 years productivity is likely to grow explosively when we develop machines, successors to our current computers, that can think in the same way that humans do. Industrial robots have already replaced many workers and as machines evolve toward full human intelligence they will gradually replace humans in all job categories. Once all types of goods and services can be produced without any human labor, productivity will essentially be infinite. Of course, this will pose serious questions about how goods and services are allocated in an economy where unemployment is 100%, how people in industrial societies share the benefits with all of humanity, how unemployed people occupy their time in meaningful and constructive ways, and how we avoid a big brother nightmare. But this productivity explosion has implications for the current day, for those policies like social security that are based on projections into the far future.
Is it likely that we will develop machines that can think and replace humans in all types of work? While neuroscientists do not completely understand how human brains work, they have made breathtaking progress in part because of the Decade of the Brain during the 1990s. Within the next 10 to 30 years our understanding of brains will tell us how to design machines that can think as humans do. And while Moore's Law about the rate of increase in circuit densities (and speeds) on computer chips is predicted to slow down in the next 10 to 15 years, progress will effectively continue in terms of the density of packing chips together, the cost of manufacturing chips, and wildcard technologies like quantum computing. There is debate among neuroscientists and computer scientists about how much computing power is required to duplicate human thinking, but we will reach that point well within the next 75 years.
The great economic lesson of the twentieth century was the importance of incentives in the struggle against scarcity. U. S. policy currently allocates most of the benefits of productivity increases to those who create the increases, as well as to consumers. However, we must also provide for the physical and spiritual needs of people whose jobs disappear because of productivity increases. Our current approach is that they find new lines of work. But during the twenty first century scarcity will finally be eliminated and there will be no new lines of work. Like generals who must be careful to avoid fighting the last war, we will need to recognize this long term shift from the problem of scarcity to the problem of allocation when everyone is unemployed. Because social security planning is based on the long term future, it needs to recognize this shift now. Current social security policy is for roughly equitable allocation of benefits; switching now to a policy of incentives for individual investment performance would be fighting the last war at just the wrong time.