(1/2) The Phenomenology of Productivity and the Great Stagnation
Part 1: Understanding the Great Stagnation
(Note: this post is part of one of two because it could not fit into a single email.)
Introduction:
In 1940, Robert Solow, a sixteen-year-old first-generation Jewish immigrant in Brooklyn, New York, got a scholarship to Harvard. He came of age in a time of cataclysmic technological output. Solow entered as an undergraduate freshman at Harvard in 1940 but would not graduate until 1947; the War held him up. Over the span of a decade, Solow had seen the United States unleash unparalleled state capacity and ultimately emerge from the Great Depression through the power of production and innovative advances in hardware. The main by-product of Solow’s worldview came in 1957 when he published his work, ‘Technical Change and the Aggregate Production Function.’ This paper gave us the Solow growth model and, by extension, the Solow Residual (also known as Total Factor Productivity). Solow’s work is used today by economists to measure the growth of productivity in society.
Today, many economists use Total Factor Productivity as a proxy to measure innovation and technology development in the economy. Moreover, productivity is the main driver for living standards in a society.
Here is the TFP formula:
Y is GDP, A is TFP, K is the economy’s total amount of capital, L is the economy’s total amount of labor, and α and β are assumptions.
The subject of my post today is not Total Factor Productivity. Instead, it is the question of the Great Stagnation. However, most conversations on the Great Stagnation begin with Total Factor Productivity and its slowing growth rate since the early 1970s.
Today, I’m going to do a few things. I’m going to insert myself into a Rorty-an conversation that has been going on for over a decade now; moreover, I’m going to be ambitious with this post.
I will do my best not to make this essay a mundane reiteration of points already made by Noah Smith (a lot of Noah Smith), Matthew Yglesias, Tyler Cowen, Jonathan Haskel, Stan Westlake, Erik Brynholfsson, Andrew McAfee, Dietrich Vollrath, Michael Strain, Matt Clancy, Nicholas Bloom, Charles I. Jones, John Van Reenen, Michael Webb and other members of the economics intelligentsia; instead, I’ll do some work to be a value add to the broader conversation towards the end of the post.
Here’s the narrative:
Since the 1970s, total factor productivity has slowed down, and phenomenological technological progress has dwindled along with it. I have three main comments on this trend:
First, we are in a lull period. This means that the technologies that will ultimately increase TFP have been in R&D for a long time.
Second, TFP does not sufficiently measure the technological progress we have made or the impacts technology has had on the lives of real people.
Third, technologies that will ultimately impact TFP already exist but still have to get through R&D before they can scale. Technologies could be brought out of this cocoon of R&D by a global crisis, Venture Capital, or big-tech go-to-market strategies.
What is the Great Stagnation?
Generally, the storyline of The Great Stagnation goes like this: there has been a slow down of Total Factor Productivity since the late 1960s or early 1970s; the year of origin varies by the economist. More importantly, the theory is that when this slow down in TFP occurred, there was a broader reduction in technology development and innovation. The TFP slowdown has unfolded alongside an increase in wealth inequality, wage stagnation, and a variety of other problems. The insinuation is that if you can solve the problem of poor technological development and, by extension, TFP, you can solve many of the social ills of our time.
Total Factor Productivity is a good metric to use when looking at technology development in society. Dietrich Vollrath states, “Growth in productivity allows you to either produce more stuff (good for growth in GDP per capita) or use fewer inputs (which is not good for growth in GDP per capita even if it is good for you or the environment). The growth rate of TFP is the better thing to focus on in asking whether stagnation matters.”
In figure 1 and figure 4, note that it looks like TFP is going up. In economics, however, you want to look at exponential growth on a logarithmic scale as opposed to linear growth. So if the linear growth rate is constant, in real terms, the growth rate is declining. Figures 2 and 3 demonstrate the actual decline.
Furthermore, Dietrich Vollrath, provides figures 2 and 4. In these figures, he uses the research from Miles S. Kimball, John G. Fernald, and Susanto Basu to correct for a discrepancy in the Total Factor Productivity assumptions. In the TFP formula, you have to make an assumption around the utilization of capacity and labor. What is clear is that even when you adjust TFP models for utilization, there has been a definitive slow down in TFP.
(Figure 1)
(Figure 2)
(Figure 3)
(Figure 4)
How Did the Stagnation Emerge?
What drove this slowdown in TFP? There’s a variety of explanations that try to diagnose this exact topic.
One explanation is from Tyler Cowen, in his book, The Great Stagnation, where he claimed that society had picked all of its low-hanging ‘intellectual fruit.’ In his view, the foundational technologies of society had already scaled in the years between 1880 and 1940. He argued that settlers had claimed all of the free lands in the US, education rates had plateaued, and ‘low-hanging fruit’ technologies were ubiquitous, “new developments include[d] electricity, electric lights, powerful motors, automobiles, airplanes, household appliances, the telephone, indoor plumbing, pharmaceuticals, mass production, the typewriter, the tape recorder, the phonograph, and radio, to name just a few, with television coming at the end of that period.”
Another explanation that some experts point to is the oil shocks of the 1970s, where engineers and innovators re-oriented their productivity away from hardware and towards software to produce goods in an energy-scarce environment. Historian Meg Jacobs in her book Panic at the Pump: The Energy Crisis and the Transformation of American Politics in the 1970s, describes how in the late 60’s the US reached its geological peak of oil production (relative to the technological capabilities of the time). As a result, the US turned towards OPEC. When the US outsourced its oil supply, those nations would occasionally jack up prices, creating oil shocks. Under these conditions, the US turned to an energy conservation policy under Nixon and Carter. While Reagan ultimately reversed this policy, and today, the US is an energy semi-autarky, the residual ‘PTSD’ remains with a general wariness about energy prices. This concern incentivizes technology that conserves energy. Note the graph below shows the slowing growth rate and ultimate plateau in energy consumption. Hardware is energy-intensive, and software is not.
(Source)
Economist Dietrich Vollrath cites a shift in demographics and that there has been a transition from a ‘goods economy’ to a ‘services economy.’ The demographics that drove TFP in the 20th century are now dragging it down in the 21st. More specifically, the Baby boom generation was so large that as they aged, the US workforce lost a significant amount of available labor. Note the chart below.
(Source)
Additionally, in the shift from ‘goods’ to ‘services’ economies, the share of GDP represented by manufacturing has gone down. To clarify, Vollrath states that the number of goods has gone up. But, the efficiencies around ‘goods’ products have increased so quickly that prices have dropped, and now hardware is a significantly smaller portion of overall economic activity. What consumers have been able to save on cheap hardware (goods), they now spend on software (services). And in the words of Robert Solow, “you can see the computer age everywhere but in the productivity statistics.”
(Source)
Total Factor Productivity vs. Visual & Lifestyle Stimuli:
The question that arises here is essential. Has technology itself slowed down, or instead have the types of discoveries made simply changed? If the needs of consumers have changed and we’re no longer focused on planes, trains, and automobiles but instead on information and communication, does that indicate that technology has not necessarily stalled but instead changed the direction of its progress? Is TFP the definitive indicator of technological advancement, and should it be relied on to form an opinion on the well-being of our societies?
Let’s double click on the ‘information and communication’ sectors.
In the world of software (services side of the economy), technology has developed consistently at an exponential rate for the last five decades. The phenomenon of Moore’s law on transistors is common knowledge and nothing short of a miracle. The range of solutions that exist today because of software as a general-purpose technology is unfathomably broad. Erik Brynjolfson and Andrew McAfee, in their book, the ‘Second Machine Age’ demonstrate just how much technological progress has occurred since the emergence of Moore’s law, “As Moore’s Law works overtime on processors, memory, sensors and many other elements of computer hardware, it does more than just make computing devices faster, cheaper, smaller and lighter. It also allows them to do things that previously seemed out of reach.”
The development of the software industry has created what Jonathan Haskel and Stan Westlake describe as the Intangibles economy. Haskel and Westlake argue alongside Dietrich Vollrath that “a significant fraction of the fall in TFP between 2000 and 2016 is explained by… the long term shift from manufacturing (where TFP growth has historically been high) to services (where TFP growth has historically been lower.)
Also, while software makes a meaningful contribution to TFP, it usually takes a while to be recognized or is sometimes missed entirely.
According to a 2018 paper by Erik Brynjolfson, “We find substantial and ongoing Productivity J-Curve effects for software in particular and computer hardware to a lesser extent. Our adjusted measure TFP is 11.3% higher than official measures at the end of 2004 and 15.9% higher than official measures at the end of 2017.” Brynjolfson is saying that TFP is missing key technological advancements!
Moreover, while TFP does not fully account for technological progress in the ‘intangibles economy,’ it is accounted for to some degree.
Brynjolfson states a slowdown in intangibles investment directly leads to a slow down in TFP. “We expect the slowdown to manifest itself in decreased TFP in which the spillovers from intangible investment - technical breakthroughs that advance an entire industry, new management methods that are widely adopted, new product designs that create whole new categories - show up… countries with the biggest slow down in the growth of intangible capital services have seen the largest slow down in TFP.” Even if intangibles are underrepresented, there is a place for tangibles and intangibles in the TFP calculus.
On the social side of this discussion, Michael Strain makes a data-driven argument that the economy, and the people who live in it, are not nearly as poorly situated as stagnation-ists would make it seem. He argues that wages have gone up, and inequality is actually going down. “The CBO found that income inequality between 1979 and 2006 increased by between 24 and 27 percent, depending on the definition of income. But things look very different between 2007 and 2016. Using market income, inequality has grown only by 2 percent. Using taxes and transfers, inequality has actually decreased by 7 percent.” Strain makes similar data-driven claims around improvements in American quality of life and upward mobility.
Dietrich Vollrath agrees with Strain. He argues that the slowdown in TFP is a sign that the primary material needs of the American people have been met to a capacity that was unimaginable just a few generations ago.
Both Vollrath and Strain are careful to qualify their arguments by saying that even though society has seen a miraculous progression, there are still a myriad of opportunities for improvement.
The key takeaway here is that the slowdown in TFP has not been accompanied by a complete deterioration of technological development overall; if that were the case, we would see it in relevant economic metrics that Strain argues have not appeared. Nonetheless, there are still ample opportunities to re-invigorate TFP and make a fairly decent situation much much better.