"Smart Machines and Long-Term Misery", Sachs & Kotlikoff;
"Are smarter machines our children’s friends? Or can they bring about a transfer from our relatively unskilled children to ourselves that leaves our children and, indeed, all our descendants – worse off? This, indeed, is the dire message of the model presented here in which smart machines substitute directly for young unskilled labor, but complement older skilled labor. The depression in the wages of the young then limits their ability to save and invest in their own skill acquisition and physical capital. This, in turn, means the next generation of young, initially unskilled workers, encounter an economy with less human and physical capital, which further drives down their wages. This process stabilizes through time, but potentially entails each newborn generation being worse off than its predecessor. We illustrate the potential for smart machines to engender long-term misery in a highly stylized two-period model. We also show that appropriate generational policy can be used to transform win-lose into win-win for all generations.
But what if the Luddites are now getting it right -- not for labor as a whole, but for unskilled labor whose wages are no longer keeping up with the average? Indeed, what if machines are getting so smart, thanks to their microprocessor brains, that they no longer need unskilled labor to operate?
Evidence of this is everywhere. Smart machines now collect our highway tolls, check us out at stores, take our blood pressure, massage our backs, give us directions, answer our phones, print our documents, transmit our messages, rock our babies, read our books, turn on our lights, shine our shoes, guard our homes, fly our planes, write our wills, teach our children, kill our enemies, and the list goes on.
Yes, technology has always been changing. But today’s change is substituting for, not complementing unskilled labor. Yesterday’s horse-drawn coaches were replaced by motorized taxis. But both required a human being with relatively little human-capital investment – a cabbie -- to drive them. Tomorrow’s cars will drive themselves, picking us up, dropping us off, and returning home all based on a few keystrokes. This will make cabbies yet another profession of the past.
Gordon (2009) also presents evidence documenting recent increases in wage inequality, including an increase in the share of wage income earned by the top 10 percent higher earners – from roughly 26 percent in 1970 to 36 percent by 2006.3 He also reports a close-to 10 percentage-point fall in labor’s share of national income since the early 1980s. This decline in labor’s overall share may also reflect accelerating growth in machine brainpower. Machines, after all, are a form of capital, and the higher income they earn based on better machine brains may show up as a return to capital, not labor income.
The Census Bureau publishes median income by age for the years 1947 to 2011.4 If we compare the median incomes of men aged 45-54 with men aged 25-34, we find that the ratio of relative income of the older cohort has risen significantly. In 1950, the income of older men was 4 percent more than their younger counterparts. In 1970, the gap was 11 percent. By 2011, the income of older men was 41 percent above the income of the younger men. For women, the trend is less apparent, with the ratio of income rising from 0.92 in 1950 to 1.15 in 1970 but then declining slightly to 1.11 in 2011. This difference may reflect that fact that men were more exposed to the downsizing of employment in manufacturing as machines replaced less-skilled workers.
As shown below, in an admittedly highly stylized life-cycle model, the general equilibrium effects of this generational redistribution can transform enhancements in machines into very bad news not just for contemporaneous young generations, but for all future generations. The model treats all young workers as unskilled agents who invest their savings in the acquisition of both skills and machines. When today’s machines get smarter, today’s young workers get poorer and save less. This, in turn, limits their own investment in themselves and in machines. The knock-on effect here is that the economy ends up in all future periods with less human and physical capital, which further depresses the first-period wages of subsequent young generations. Although the skilled wage premium and the return to capital rises, the net impact of smartening up today’s machines is a reduction in the lifetime wellbeing of today’s and tomorrow’s new generations. In short, better machines can spell universal and permanent misery for our progeny unless the government uses generational policy to transform win-lose into win-win.
We see that a rise in machine productivity reduces the unskilled wage if εML > εSN/θ. Immiserizing productivity is more likely if:
- Substitutability of machines and unskilled labor is high (εML large)
- Substitutability of intermediate goods and skilled labor is low (εSN small)
- The share of skilled labor in final output is high (θ high)
Note that immiserizing productivity is not possible for a Cobb-Douglas production function or indeed for any aggregate CES production function in which εML = εSN. In those cases, a rise in machine productivity necessarily raises the wage of unskilled labor.
Our point can be simply summarized. Suppose that an innovation in machine technology (e.g. improved software) raises machine productivity in a manner that indeed reduces the marginal productivity of low-skilled workers while raising the marginal productivity of high-skilled workers. This not only increases the income gap between skilled and unskilled workers, but also has a generational effect, raising the incomes of the older generation while lowering the income of the young. This effect occurs because the old have accumulated physical and human capital, while the young are endowed with unskilled labor. The generational redistribution has a knock-on effect on national saving. Income shifts from young savers to older dis-savers, thereby depressing the national saving rate and the future stock of capital."
"Are smarter machines our children’s friends? Or can they bring about a transfer from our relatively unskilled children to ourselves that leaves our children and, indeed, all our descendants – worse off? This, indeed, is the dire message of the model presented here in which smart machines substitute directly for young unskilled labor, but complement older skilled labor. The depression in the wages of the young then limits their ability to save and invest in their own skill acquisition and physical capital. This, in turn, means the next generation of young, initially unskilled workers, encounter an economy with less human and physical capital, which further drives down their wages. This process stabilizes through time, but potentially entails each newborn generation being worse off than its predecessor. We illustrate the potential for smart machines to engender long-term misery in a highly stylized two-period model. We also show that appropriate generational policy can be used to transform win-lose into win-win for all generations.
But what if the Luddites are now getting it right -- not for labor as a whole, but for unskilled labor whose wages are no longer keeping up with the average? Indeed, what if machines are getting so smart, thanks to their microprocessor brains, that they no longer need unskilled labor to operate?
Evidence of this is everywhere. Smart machines now collect our highway tolls, check us out at stores, take our blood pressure, massage our backs, give us directions, answer our phones, print our documents, transmit our messages, rock our babies, read our books, turn on our lights, shine our shoes, guard our homes, fly our planes, write our wills, teach our children, kill our enemies, and the list goes on.
Yes, technology has always been changing. But today’s change is substituting for, not complementing unskilled labor. Yesterday’s horse-drawn coaches were replaced by motorized taxis. But both required a human being with relatively little human-capital investment – a cabbie -- to drive them. Tomorrow’s cars will drive themselves, picking us up, dropping us off, and returning home all based on a few keystrokes. This will make cabbies yet another profession of the past.
Gordon (2009) also presents evidence documenting recent increases in wage inequality, including an increase in the share of wage income earned by the top 10 percent higher earners – from roughly 26 percent in 1970 to 36 percent by 2006.3 He also reports a close-to 10 percentage-point fall in labor’s share of national income since the early 1980s. This decline in labor’s overall share may also reflect accelerating growth in machine brainpower. Machines, after all, are a form of capital, and the higher income they earn based on better machine brains may show up as a return to capital, not labor income.
The Census Bureau publishes median income by age for the years 1947 to 2011.4 If we compare the median incomes of men aged 45-54 with men aged 25-34, we find that the ratio of relative income of the older cohort has risen significantly. In 1950, the income of older men was 4 percent more than their younger counterparts. In 1970, the gap was 11 percent. By 2011, the income of older men was 41 percent above the income of the younger men. For women, the trend is less apparent, with the ratio of income rising from 0.92 in 1950 to 1.15 in 1970 but then declining slightly to 1.11 in 2011. This difference may reflect that fact that men were more exposed to the downsizing of employment in manufacturing as machines replaced less-skilled workers.
As shown below, in an admittedly highly stylized life-cycle model, the general equilibrium effects of this generational redistribution can transform enhancements in machines into very bad news not just for contemporaneous young generations, but for all future generations. The model treats all young workers as unskilled agents who invest their savings in the acquisition of both skills and machines. When today’s machines get smarter, today’s young workers get poorer and save less. This, in turn, limits their own investment in themselves and in machines. The knock-on effect here is that the economy ends up in all future periods with less human and physical capital, which further depresses the first-period wages of subsequent young generations. Although the skilled wage premium and the return to capital rises, the net impact of smartening up today’s machines is a reduction in the lifetime wellbeing of today’s and tomorrow’s new generations. In short, better machines can spell universal and permanent misery for our progeny unless the government uses generational policy to transform win-lose into win-win.
We see that a rise in machine productivity reduces the unskilled wage if εML > εSN/θ. Immiserizing productivity is more likely if:
- Substitutability of machines and unskilled labor is high (εML large)
- Substitutability of intermediate goods and skilled labor is low (εSN small)
- The share of skilled labor in final output is high (θ high)
Note that immiserizing productivity is not possible for a Cobb-Douglas production function or indeed for any aggregate CES production function in which εML = εSN. In those cases, a rise in machine productivity necessarily raises the wage of unskilled labor.
Our point can be simply summarized. Suppose that an innovation in machine technology (e.g. improved software) raises machine productivity in a manner that indeed reduces the marginal productivity of low-skilled workers while raising the marginal productivity of high-skilled workers. This not only increases the income gap between skilled and unskilled workers, but also has a generational effect, raising the incomes of the older generation while lowering the income of the young. This effect occurs because the old have accumulated physical and human capital, while the young are endowed with unskilled labor. The generational redistribution has a knock-on effect on national saving. Income shifts from young savers to older dis-savers, thereby depressing the national saving rate and the future stock of capital."
What about families? Which leads to the next question: What about class?
Is this an attempt to justify wealth redistribution without mentioning social class?Jan 21, 2013
