This is my final note in this series, it is on jobs and automation. Advances in automation drive increased productivity, and result in job losses. In fact, contrary to some claims by politicians, manufacturing job losses in the past half-century were caused primarily by automation and not by “unfair trade”. Automation brings significant benefits, but it causes significant social dislocation we must address both locally and globally.
Why automation?
All goods producers (but for a few who, because of specific circumstances and/or beliefs, are exceptions to the general trend) in agriculture, mining, and manufacturing replace human labor with machines whenever the state of technology makes this cost-effectively possible. Furthermore, when such replacement is technologically possible, systematic engineering improvements will make it cost-effective. In farming, the multi-function integrated “combine” replaced horse and man, and now combines yield to GPS-controlled driverless planting/harvesting machinery. In mining, blowing off the top of mountains allowed scooping coal directly onto trucks that now morph into their driverless versions. In manufacturing, hard automation first replaced manual labor, now flexibly programmable “smart factories” are coming.
Why is this trend of replacing people with machines so strong and unstoppable? Producers want to minimize costs and optimize quality. In fact, any quality failure adds to costs, thus, lack of quality is a cost component. A quality failure in the plant, i.e. a reject, always means some combination of losses of labor invested, materials used, and machine time deployed. A quality failure in the field is typically much more expensive than those in-plant; field failures incur significant service and possibly legal costs too. People make mistakes and require breaks. Well-maintained machines do not make mistakes and can operate continuously around the clock.
Furthermore, the very nature of many advanced products would make their manual fabrication inconceivable. Thus, automation is and/or can be a cost reducer, a quality improver, and a new product enabler; it is one of two fundamental drivers of economic growth, and the other one is trade.
Evolution
Machines in early factories were mainly machine tools (e.g., lathes, drills). They were powered by a central steam engine that drove the main drive shaft mounted above the shop floor close to the ceiling. The machines were connected to the drive shaft by belt and pulley systems. After the invention of the electric motor, machine tools could be driven directly by such motors. Around 1900 this led to the development of assembly lines. New factories could be organized into lines; work stations along the line could be equipped with these independently driven tools. The principal purpose of the line was to convey the product that was then worked on progressively by operators located at the work stations along the line. Later, in the mid-1900s, “hard automation” was introduced to process mass-produced parts and sub-assemblies. For hard automation, each part and sub-assembly required the design and construction of a unique piece of equipment dedicated to the specific part and processing task. The loading of parts was typically done either manually into appropriate jigs, or the parts arrived from the parts suppliers in “well-defined-geometry” packages. The automated machine then performed pick-and-place operations and the subsequent processing of the precision positioned parts, for example, soldering of electronic parts or welding mechanical components. Hard automation needed to be augmented by the manual dexterity and vision of human operators. But the machines could be fast and precise.
The first robots introduced in the second half of the 20th century had some tactile sensing and dexterity, but they lacked vision. They were primarily transfer devices to move parts and assemblies from one place to another; they required less precise pre-positioning of parts, had more dexterity, and were more easily adapted to handling different parts than were the transfer systems in hard automation. Their movements could be changed by reprogramming, and a variety of pick-up devices (vacuum or mechanical) and tools could be attached to the dexterous hands. Lacking vision, their movements were mostly by dead reckoning. By about 2010 machine vision became well developed and robots could begin to “look”, though only in well-defined environments, e.g., recognizing objects having preprogrammed shapes and then performing preprogrammed operations on such objects.
The Future
Artificial intelligence (AI), including artificial neural networks (ANN), is now the subject of major development efforts. Like human vision, where the eye is the sensor, but the brain does the seeing, machine vision augmented by AI enables machines to not only look but truly “see”. This will lead to vision-based control of flexible robotic operations; such machines are expected to be able to operate autonomously in a wide variety of environments. Here we are not talking only about manufacturing, but about a wide range of activities that till recently required human operators. Think about self-driving cars and trucks operating on public roads. Think about automated manufacturing lines feeding automated warehouses delivering goods to consumers via drones or autonomous vehicles. And brainy machines are not restricted to working with goods only; they are well suited to join the service industries too. Hospitals already utilize robotic devices for drug delivery to nursing stations; further impacts on jobs by automation are likely in hospitals, home care, cleaning, and the list goes on. The list of areas where AI-augmented, seeing, dexterous, and mobile machines are likely to replace humans, is limitless.
How many jobs will be eliminated by these smart machines that are not yet fully developed, but for sure are coming? Obviously, it will depend on their cost and performance. The question is not whether but when? One estimate I read says that by the early-2030s, 38% of the jobs in the US, 35% in Germany, 30% in the UK, and 21% in Japan are at risk to be replaced by automation. (http://www.pwc.co.uk/economic-services/ukeo/pwcukeo-section-4-automation-march-2017-v2.pdf) While such predictions are never precise, and no doubt in the evolving economies new jobs will be created, some of them directly associated with the growth in automation. Two trends are clear. First: many jobs will disappear; second: the most affected will be workers with low education.
Consequences
The cost of goods historically shared three basic components: the costs of materials, labor, and capital. All these costs imply incomes shared elsewhere. As automation replaces labor, the share of capital goes up and the income of those who provide the capital goes up, while the share of labor’s income goes down. People who lost their jobs have no income; people who provide the capital to buy robots, and those who provide their intellectual capital to develop them, gain income. There follows a growing inequality that must be addressed.
Our highly-automated farms produce enough food to feed everyone; obesity is a much bigger issue in the USA than undernourishment. The automated factories can satisfy all consumer needs. But, we face a growing inequality. Even though the GDP per capita between 1970 and 2015 more than doubled, the earnings of people at the lower 40% income level increased by only 13%; between 2000 and 2015 the GDP per capita increased by 13 %, and the income at the 95th percentile increased by 7%, while the income at the 20th, 40th, and 50th percentiles decreased. (Data sources: census.gov and worldbank.org on the web.)
Growing inequality is a major, and growing, socio-economic and political issue. The outcome of the last US presidential election was decided by people in the rust belt who felt that they were left behind by the evil forces of globalization and automation. While cheap rhetoric may win elections, it does not change the fact that a prosperous and peaceful future will have more global trade and more automation. Serious effort is needed to address the real problem of inequality.
Actions needed
I believe there are three fundamental elements to the solution: reduce the standard hours worked, increase education opportunities, and strengthen the social safety net. An adjustment of the standard workweek is long overdue; the first 40-hour workweek was introduced over a hundred years ago; since then, US GDP per capita grew more than ten-fold. Education drives intellectual capital, lifetime earnings correlate with the level of education, and unemployment anti-correlates. Educated people usually keep learning throughout their lives, a trait much valued in the fast-changing modern economy. A revised and strengthened social safety net is essential, both for ethical and pragmatic political reasons. Going forward, there is virtually no need for purely manual labor. Not everyone will be employable. To lead a complete and satisfactory life, everybody needs food, housing, healthcare, safety, education, and some other amenities. To avoid catastrophic conflicts and preserve our democracy, we have an obligation to take care of all of us, even those who are unable to earn a living. We cannot, and should not desire to stop the beneficial forces of globalization and automation on productivity growth. But we all must share the benefits derived from it; thus, we should, and must, take care of all of us; especially those of us who are unable to earn a living in our automated world.
Much of this is contrary to traditional American thinking. But, it is time to update our traditions. The founding fathers lived in a totally different world from ours: one that was much poorer, much more isolated, and not automated. We are now, and going forward: rich, integrated, and automated. We need to work less and learn and think more. To correct one major labor mistake handed down by the founding fathers, we already had one bloody civil war; to prevent and avoid the next one, we must now take civil action.
Addendum to my blog series on
Postscript. While much of this discussion was America-focused, most ideas presented here are applicable to countries enjoying advanced economies elsewhere, e.g., to the EU. True, Europeans do not carry the baggage of our “founding fathers”, they face other serious baggage. For example, the very history and geographic location of the USA make it much more effective in integrating immigrants than are the European nations facing the massive 21st-century waves of African and Asian dispossessed. The force of immigration only adds to those of trade and automation in driving the disaffected toward nationalistic isolationism, which in the short term is politically disruptive, and in the long term destructive.
Manufacturing, Trade, and Jobs
Acknowledgment: this Addendum was prompted by exchanges with interested readers of my blog series on “Manufacturing, Trade, and Jobs” and was added subsequent to the original posting of the series. Their comments focused on two broad subjects: 1) the cause of the ongoing trend of growing inequality, and 2) trends and fluctuations in the US GDP/capita growth. I thank them for their suggestions and insights.
One finds various published explanations for the worldwide growing inequality. Some suggest that “headwinds” and/or “tailwinds” that help and/or block an individual’s progress from early on may be the determining factors; thus providing “tailwind” assistance through social programs could provide a solution to the social problem of inequality. (Sendhil Mullainathan: “Taking a Different Approach to Inequality”, NYTimes, Sun. Apr. 30, 2017, p. BU5.) While such assistance is a welcome idea, and it could make a positive difference in the lives of many individuals, it would not eliminate the root cause. The root cause, in my opinion, is that automation is providing an economic benefit by cost-effectively replacing labor in progressively broader segments of the economy. Specifically, both in goods production and services, the basic cost elements are materials, labor, and depreciation of the cost of the tools deployed in the activity. As technology improves, automation replaces labor, and depreciation accounts for a progressively larger fraction of the activity’s basic cost. Depreciation is essentially the repayment with interest of the capital utilized in establishing and facilitating the activity. A secondary factor is that capital grows by compounding over time; while labor benefits from experience, and may compound over a practitioner’s life, those who do reasonably well must learn new tricks all the time to update their skills. Thus, in our current social-economic arrangement, the cause is that capital is being substituted for labor, and the effect is inequality. - - I do want to emphasize that I believe that automation is serving the common good and what we need to address is the distribution of its benefits so that inequality is mitigated and the explosive social tension driven by the growing inequality is resolved ethically and pragmatically.
Much of my thinking in these matters is based on a conviction that the US economy is basically sound, and is showing a secular growth trend; we have not reached the end of history. It is reasonable to question that conviction, especially given that on April 18, 2017, the BEA published that the real GDP in Q1 2017 grew at an annual rate of 0.7 %; this is the lowest growth in the past three years. (https://www.bea.gov/newsreleases/national/gdp/gdpnewsrelease.htm)
To address the growth question, I looked at 50 years of data, both in terms of growth and fluctuations to identify factors possibly hidden under the longer-term trend. From 1966 to 2015 the economy as measured by GDP/capita grew about two-and-a-half fold (during the same period the GDP grew about four-fold since the population increased more than one-and-a-half fold). Each decade in this half-century showed growth, though each had one or more years of contraction (negative growth). The 1976-85 period showed the highest aggregate growth, the highest single-year growth, and the highest variability as measured both by absolute deviation and range. The last decade had the lowest growth and the worst recession since the Great Depression. The next-to-last decade showed the second-highest growth and it had the lowest downturn year in the half-century. Is there an obvious trend, other than aggregate growth?
I don’t think so, though there are others who think we have reached the end of growth and that innovation is dead. (https://www.ted.com/talks/robert_gordon_the_death_of_innovation_the_end_of_growth/transcript?language=en). Clearly, growth requires continuous innovation. In the advanced world, mankind’s basic needs of food, shelter, clothing, and other basic amenities are now satisfied. Future innovators may need to work harder to identify needs they may satisfy with their inventions. At the same time, innovators now are much better trained, have better tools, and may face bigger issues than their forebears did. The environment, clean energy, and clean transportation are examples of major innovation and growth opportunities, also health and education services are likely to grow and be impacted by advances in technology. Maybe, in the advanced world, future innovation and economic growth will come from satisfying communal, rather than individual, needs. In any case, I expect innovation and economic growth to continue.
In closing, I present some measures of the problems the US economy is facing. These problems cause the growing conflict between those who feel they have been left out and the rest. (The “rest” includes much of the political and economic establishment as seen by the “left outs”.) The available income-distribution data shows how the majority has not fully shared in the benefits of our growing economy. In real (inflation-adjusted) terms, the Real Median Household Income peaked in 1999. Thus, in the past more than a quarter of a century, most Americans have experienced a declining living standard, while the economy has been growing. In fact, the median household lost 2% income, while the economy per capita gained 18%. Furthermore, even though between 1993 and 2015 the median still increased, most of that increase occurred before 1999. During this close to a quarter-century from 1993 to 2015 period, even though the nation’s per capita GDP went up 39%, the median household’s income increased only by 12%. Thus, the middle-class family was not benefiting proportionately from the national economy’s continued success. Most of the gain went to the upper-income groups. The most accepted measure of economic inequality is the Gini index; a Gini index of zero (0) implies that all members of the community share equally the benefits, while an index of one (1) implies total inequality where an infinitesimal minority takes all the benefits. Between 1993 and 2015 the money income Gini index increased by 5.5%.
To summarize: the economy has been growing and is likely to continue to grow. But, endemic factors associated with the progressive transition from labor-intensive to automation-intensive activities across most of the economy, have caused economic inequality to grow. A rebalancing is required to allow all to share the benefits derived from technological progress.
No comments:
Post a Comment