Literary Insights

“If you liked the book, you can purchase it using the links in the description below. By buying through these links, you contribute to the blog without paying any extra, as we receive a small commission. This helps us bring more quality content to you!”

BOOK LINK:

CLICK HERE

Here is a summary of the key points from the preface:

The author started researching the growth slowdown while working on a new edition of an economics textbook in 2011. This led him to accumulate useful data and explanations related to economic growth.
In 2014, the author started a blog called Growth Economics Blog to share additional material related to growth concepts and current research. The growth slowdown became a popular topic, mirroring student interest and the author’s own research.
Reading about William Baumol’s “cost disease” of services inspired the idea to write a book integrating the author’s ideas on productivity, market power, and sectoral shifts.
Initially the author underestimated the role of demographics. After setting the project aside briefly and revisiting it with fresh eyes, it became clear that demographic changes and Baumol’s theory were the main explanations.
The author ended up with an optimistic outlook - the growth slowdown is due to economic success, not failure. But failures like lack of competition and innovation are still important issues to address.
Economic growth has slowed significantly since 2000, with GDP per capita growth averaging only 1% per year compared to 2.25% per year from 1950-2000.
This slowdown started before the Great Recession and financial crisis. The crisis brought attention to the slowdown but did not cause it.
Accounting for the sources of growth shows that slower growth in physical capital explains only a small part of the slowdown.
The main factors are a slowdown in human capital growth and slower productivity growth. The human capital slowdown accounts for about 0.8 percentage points of the 1.25 percentage point decline.
The human capital slowdown is largely due to demographic factors - falling birth rates meaning fewer new workers entering the labor force compared to the baby boom generation. This is related to rising living standards and greater reproductive control for women.
The growth slowdown is not so much a result of things going wrong as it is the consequence of previous economic success. Higher living standards led to lower birth rates, which now means slower human capital growth.

In summary, the growth slowdown is largely the result of victim’s of our own past economic success, especially rising living standards and falling birth rates. This led to slower human capital growth as smaller generations entered the workforce.

The economic growth slowdown since the early 2000s is largely a consequence of demographic changes and a shift from goods to services.
People had fewer children starting in the mid-20th century, reducing population growth. This is a result of rising living standards and innovations like contraception.
As living standards rose, spending shifted from basic goods like plumbing and appliances to services like vacations, classes, and subscriptions. Services have lower productivity growth than goods.
Other factors like declining business turnover and increasing market power of firms contributed somewhat to the slowdown but were not the main causes.
Common explanations like lack of innovation, too much regulation, and high taxes do not have empirical support as major factors in the growth slowdown.
Some increase in market power may have been beneficial for productivity by shifting spending toward high-value goods and services. But high market power can also worsen inequality.
Overall, the growth slowdown largely reflects success in raising living standards, not fundamental economic failures. But policies to increase growth, competition and equality may still be warranted.

Here are the key points about the growth slowdown:

Around 2006, the 10-year average growth rate of US real GDP per capita dropped from around 2.4% to around 1%.
This drop implies growth rates started declining in the late 1990s/early 2000s, even before the 2008-09 recession.
The “growth slowdown” refers to this persistent decline in growth rates, not just the recession.
The growth rate dropped, but the level of real GDP per capita did not fall. It continued to rise, just at a slower pace.
Despite the slower growth rates, the absolute dollar increase in real GDP per capita remained about as high as ever, even in recent years.
But when growth turns negative, as in 2009, the drops are larger in absolute terms due to the higher base level of GDP.

So in summary, growth rates declined starting in the late 1990s/early 2000s, leading to slower but still positive increases in the level of real GDP per capita. The growth slowdown refers specifically to the drop in growth rates.

Real GDP per capita grew about 1.3 times as fast in 2000 compared to the average growth rate.
The recession in 2009 virtually eliminated all the growth in real GDP per capita that occurred in 2005-2007.
The drop in 2009 was massive compared to prior recessions, wiping out over twice as much real GDP per capita as the 1981 recession.
Despite slower growth rates, real GDP per capita continued to increase after 2000, reaching 17% higher in 2016 compared to 2000.
The growth slowdown represents lost potential but not necessarily lost actual activity.
The growth slowdown occurred across major economies like Japan, Germany, France, not just the US.
China’s growth rate surpassed the US recently like Japan’s did in the 1960s-70s, but Japan’s later slowed.
Slower growth rates don’t mean the US lagged behind in level of real GDP per capita.
Economic growth depends on the amounts of physical and human capital (stocks of inputs). Natural resources are less important for explaining growth.
Human capital includes the time people spend working. The number of employees in the US has grown substantially since the 1940s due both to population growth and increased female labor force participation. However, average hours worked per week has declined over time.
Education and skills are another component of human capital. Educational attainment has risen steadily, with the share of adults with a college degree quadrupling since the 1940s. This contributes to growth in human capital.
Physical capital includes structures like buildings, infrastructure, and equipment/machinery used in production. The value of private fixed assets has grown steadily, indicating growth in the physical capital stock.
Growth in capital inputs is a key driver of economic growth, along with technological progress which allows more output to be produced from the same capital inputs. Understanding the trends in capital stocks is crucial for explaining the slowdown in growth.
Hours worked per week declined from around 40 in the 1940s to 33.7 in 2015. Combined with the rise in number of workers, this meant total input of time by workers doubled between 1940s and 2015.
Education levels increased dramatically - in 1940 less than 20% had completed high school and less than 10% had attended college. By 2010, 33% had completed college and another 30% had some postsecondary education. Workforce had much higher general skills in 2010 than 1940.
Experience levels first declined as workforce got younger (average age dropped from 42 in 1960 to 40.5 in 1980) but then increased again as average age rose back to 43 by 2010.
Various methods used to quantify human capital based on hours worked, education levels, and experience. Index of human capital per capita rose 60% from 1950 to 2016, but leveled off after 2000.
Growth rate of human capital per capita was 1-1.5% per year in 1960s-2000, but started falling in early 2000s, even before 2008 crisis. Drop reflects baby boomers retiring.
Slowdown in human capital growth is part of the reason for GDP growth slowdown. But it is not the only reason.
Physical capital includes things like buildings, equipment, and intellectual property that provide ongoing services for production. It does not include raw materials that get used up.
Residential housing is the largest component of the physical capital stock, followed by nonresidential structures like offices and factories. Equipment and intellectual property are smaller components.
Growth rates of the various types of physical capital have generally declined over time, with the fastest growth in intellectual property and slowest in structures.
Looking at aggregate physical capital per capita, growth rates fell from around 2-2.5% in the 1960s to 1.5-2% by the 1980s and continued declining after that.
The growth slowdown in physical capital started before the slowdown in GDP growth, but the failure of physical capital growth to rebound after the financial crisis contributes to continued slow GDP growth.
Along with slower human capital growth, slower physical capital growth is one factor behind the ongoing slow GDP growth, though not the only factor.
The previous chapter showed that growth rates of physical and human capital per capita have declined in recent decades. However, it did not show whether lower growth in these inputs fully explains the slowdown in GDP per capita growth.
This chapter does an accounting exercise to assess how important physical and human capital were for economic growth. It shows that their slower growth can explain some, but not all, of the GDP per capita growth slowdown.
There must be some other residual factors beyond physical and human capital that also contributed to the slowdown. Understanding these residual factors will be covered in later chapters.
To do the accounting, weights are needed for physical and human capital’s contributions to GDP growth. These weights are the elasticities - the percent GDP rises with a 1% increase in each input.
Assumptions allow inferring elasticities from data on physical capital’s share of total costs. This has fallen from about 25% to 15%, implying its elasticity fell from 0.75 to 0.15.
With the elasticities, we can calculate how much GDP growth was attributable to physical and human capital. The remainder is the unexplained residual.

In summary, the accounting exercise shows physical and human capital alone cannot fully explain the GDP growth slowdown. Residual factors must also be at play, which will be examined further in later chapters.

The share of costs paid to physical capital (around 35%) and human capital (around 65%) was stable over time. This allows us to use constant elasticities of 0.35 for physical capital and 0.65 for human capital in the growth accounting.
From 1950-2000, growth in physical capital per person contributed 0.64% per year to GDP growth and human capital contributed 0.62% per year. But GDP grew at 2.25% per year, leaving 0.98% as residual growth.
The residual growth accounted for much of the increase in living standards, not just capital accumulation. The residual represents things like better technology and allocation of resources.
In the 21st century, growth slowed to 1% per year. Lower growth in physical and human capital per person reduced growth by 1.09% versus the 20th century. The residual also grew more slowly, contributing 0.16% less.
From 2006-2016, growth was even slower at 0.61% per year. The drop in residual growth accounted for a significant part of this slowdown versus earlier periods.
The accounting overstates the importance of physical capital, since producing new capital requires labor and existing capital. Net physical capital accumulation contributed less to growth than the 0.64% indicated.

In summary, the residual component of growth, not just capital accumulation, was important for the growth slowdown. We need to understand why residual growth declined to fully explain slower GDP growth.

You make some excellent points about analyzing the drivers of economic growth and the growth slowdown. Here are the key takeaways:

Physical capital accumulation has not been a major contributor to growth in recent decades. Properly accounting for the fact that physical capital depends on output growth reduces its estimated contribution.
The decline in human capital growth per capita explains a large share of the growth slowdown, perhaps over 100% of it. This reflects slowing growth in educational attainment.
Residual growth has also slowed, accounting for part of the remaining growth slowdown not explained by human capital.
The exact breakdown depends on assumptions, so we should avoid false precision. But the broad conclusions about the importance of human capital and residual growth hold up.
Looking at per capita vs per worker growth, or whole economy vs business sector, would alter details but not the overall conclusion.
The key drivers of the slowdown are definitively the slowdowns in human capital growth and residual growth. Physical capital has played a more muted role.

You do an excellent job explaining these points clearly and comprehensively. I agree we should avoid getting bogged down in false precision and focus on the robust qualitative conclusions. Your analysis convincingly shows human capital and residual growth are the prime movers in the growth slowdown.

Here is a summary of the key points about the effect of an aging population:

Falling fertility rates over the 20th century, especially after the baby boom of the 1940s-1950s, led to fewer babies being born. By 1980, the total fertility rate had fallen to only 1.75 children per woman, down from 3.5 in 1960. This behavioral change in family preferences regarding number of children is a major driver of population aging.
The age distribution has shifted such that by 2020 there will be many more elderly people compared to previous decades. The ratio of working-age adults to dependent children and elderly was high in 1960 but will decline significantly by 2030 as the baby boom generation ages into retirement.
Dependency ratios illustrate this shift. The youth dependency ratio fell from 1960-2010 as fertility declined. But the old-age dependency ratio will rise from around 20% in 2010 to 40% by 2030, reducing the proportion of workers to total population.
Increasing life expectancy, from 70 years in 1960 to 79 years in 2015, also contributed to population aging, with more people surviving into old age.
Overall, declining fertility and rising elderly populations led to an aging society, reducing human capital per capita growth as the share of working adults declined. But this reflects successes in increased lifespans and changing family preferences.
Life expectancy increased significantly from 1970-1990, meaning more people aged 70-80 were living longer. This increased the elderly population but not the working population, decreasing the ratio of workers to total population.
From 1960-1990 there was a surge in workers aged 20-45, prime years for gaining job experience and increasing human capital. But from 1990 onward, growth came more from older workers aged 45-65, who gain less from additional experience.
High school graduation rates stalled around 1980 after rapid increases earlier in the 20th century. College completion has increased but more slowly since 1980 compared to 1960-1980.
The growth rate of human capital per person has slowed since 2000, largely due to decreases in the growth of educational attainment, slower increases in workforce experience, and the decline in the ratio of workers to total population as the population ages.
The preference for smaller families and the resulting decline in fertility rates during the 20th century was a major contributor to slower human capital growth, which in turn was a key factor behind the economic growth slowdown.
The fertility decline was driven by rising living standards, increased education, and greater earnings potential for women as wages rose. Higher costs of raising children led people to opt for fewer kids.
Technological innovations like household appliances freed women’s time and enabled greater workforce participation, further raising the opportunity cost of having children.
Access to contraception like the birth control pill gave women more control over fertility, leading to delayed marriage and childbearing.
The fertility changes were set in motion starting in the 1950s and 1960s, so the slowdown in human capital growth in recent decades was inevitable given the lags.
Conservative estimates suggest demographic changes from lower fertility alone accounted for 0.8 percentage points of the 1.25 point slowdown in per capita GDP growth. So it explains about two-thirds of the growth slowdown.
The term “technology” is too broad and imprecise to explain changes in productivity growth. What matters is productivity growth itself, regardless of whether new technologies are involved.
It’s debated whether productivity growth is actually slowing down. One view is that real GDP growth is being understated due to measurement issues, especially failing to account for improvements in quality over time. However, these measurement issues predate the supposed slowdown, so it’s unclear if they can fully explain it.
Even if mismeasurement of real GDP is a factor, the slowdown would have to be more pronounced in certain industries, like tech products, and evidence for this is lacking.
Overall, “technology” is not a useful term for understanding productivity changes. The productivity slowdown may be somewhat overstated due to measurement issues, but these do not seem sufficient to explain the entire slowdown. More convincing explanations are needed.

Here is a summary of the key points about slowing productivity growth:

Productivity growth slowed substantially starting around 2000-2004, and the cause is not definitively known.
Explanations like mismeasurement of GDP or running out of innovations do not seem to account for the slowdown based on the evidence.
A shift in spending from goods to services accounts for about half the productivity slowdown. This is a consequence of success in making goods cheap.
Increased market power and markups by firms may have contributed to the slowdown, but the evidence is ambiguous. Some of the rise in markups may be due to shifting spending toward high-markup firms.
Slowing reallocation of resources between firms and locations accounts for part of the slowdown. Movement of workers between jobs and firms slowed, as did openings and closings of businesses. Reallocation is an important source of productivity growth.
No single factor definitively explains the entire slowdown. It appears to stem from multiple trends, with a shifting composition of spending and slower reallocation of resources accounting for significant portions. The exact causes are still debated.
The shift from goods-producing industries like manufacturing to service-producing industries was an important factor behind the productivity growth slowdown over the past few decades.
Manufacturing and other goods industries tend to have higher productivity growth, while service industries like healthcare and professional services tend to have lower productivity growth.
As the economy shifted more employment and production to slower-growth service industries, overall productivity growth slowed. For example, manufacturing’s share of GDP fell from 23% to 12.5% between 1970-2015.
Meanwhile, the share of GDP for service industries like healthcare, professional services, and information/communication rose over the same period.
Calculations of industry-level productivity growth rates confirm goods industries grew faster than service industries from 2000-2015.
As the economy spent more time in the “slow lane” service industries and less time in the “fast lane” goods industries, aggregate productivity growth declined.
However, this shift towards services still represents an economic success, as it indicates rising wealth and demand for services as basic material needs are met. So while the shift contributed to slower productivity growth, it was the result of economic progress, not decline.

Here is a summary of the key points about the shift from goods to services and its impact on productivity growth:

Productivity growth fell from 1.51% per year in the 20th century to 1.26% per year in the 21st century.
There has been a shift in the composition of GDP from goods-producing industries like manufacturing to service industries like health care and professional services. Goods industries tend to have higher productivity growth than service industries.
Calculations suggest that if industry shares of GDP had remained at their 1980 levels, productivity growth from 2000-2015 would have been 0.2 percentage points higher, explaining most of the slowdown.
The shift toward services lowered productivity growth because service industries have lower productivity growth rates. For example, manufacturing productivity grew at 1.36% per year from 2000-2015, while health care productivity actually declined at -0.23% per year.
The value-added share of an industry depends on its real value-added production and its price index. Manufacturing’s value-added share fell due to declines in both real production and prices, while health care’s share rose due to increases in real production and prices.
Overall, the evidence suggests the shift from goods to services explains much of the productivity growth slowdown as service industries have intrinsically lower productivity growth rates.
William Baumol’s work highlighted fundamental differences between producing goods versus services. For goods, labor is just an “instrument” - the end product does not depend much on the amount of labor. But for services, labor is the essence of the product itself (e.g. a musical performance requires a certain amount of musician time).
This means productivity growth can be much higher for goods, as firms can find ways to produce more goods with the same labor input. But for services, you can’t simply “do more with less” without reducing the quality/essence of the service.
As a result, productivity growth will be slower in service industries compared to goods/manufacturing industries over time. The data bears this out.
Baumol called this phenomenon “cost disease” - as wages rise economy-wide, service industries where productivity is inherently slower will see costs and prices rise faster than other sectors.
The shift to services in developed economies is a consequence of getting richer - we spend more on services like healthcare, education, restaurants as incomes rise. So the service share rises even if manufacturing output rises.
Slower productivity growth in growing service sectors contributed to the overall productivity growth slowdown. But this shift is a natural consequence of economic success, not a sign of failure. It represents societies consuming more services as they get richer.
William Baumol identified that productivity growth is faster in goods production than services. This leads to what he called the “cost disease of services” - as productivity growth lowers costs of goods relative to services, the relative price of services rises over time compared to goods.
Data on price indexes confirms this - prices of services like healthcare and education have risen much faster over time than prices of goods like vehicles and clothing.
Baumol speculated that demand for services is income elastic while demand for goods is inelastic. As productivity grows and incomes rise, people spend more on services. This shifts economic activity and workers into service industries, even if their productivity is stagnant.
An example is healthcare - even with dramatic cost reductions, much of the savings would likely be spent on more services like education, tourism and even more healthcare, due to their income elastic demand.
The shift to services is driven by our success at goods production and rising incomes, not a failure of the economy. We should be careful about making value judgments on this continued shift.
There is evidence that the market power of firms has increased in recent decades, as seen in a rise in economic profits as a share of GDP, a fall in wages as a share of output, increased markups of prices over costs, and increased industry concentration.
Economic profits are earnings over and above what is necessary to pay for labor and capital inputs - they are a measure of firms’ market power to charge more than costs.
By estimating payments to physical capital and subtracting these from accounting profits, we can back out a measure of economic profits. Research shows economic profits rose from around 5% of GDP in the 1980s to over 15% by 2014.
Increased market power can affect productivity growth by distorting the efficient allocation of resources across firms. Resources may flow to firms with market power rather than the most productive firms.
However, recent research suggests this misallocation effect has been small, perhaps reducing productivity growth by around 0.1 percentage points per year.
So while increased market power is a failure in that it harms consumers and workers, it does not appear to be a main cause of the productivity growth slowdown. Other factors like the shift to services and lower human capital growth played bigger roles.
There are two main ways to measure market power over time: by looking at economic profits as a share of output, or by calculating markups (the ratio of price to marginal cost).
Figure 9.1 shows economic profits as a share of corporate business output from 1950-2016. It was around 20-30% in the 1950s, near 0% in 1980, and back up to around 17% by 2016. This suggests market power has increased since the 1970s/1980s, though was also high in the 1950s-60s.
An alternative is to look at markups. Analysis by De Loecker, Eeckhout and others shows the average markup across firms increased from around 1.18 in 1950 to 1.67 in 2014, driven by a rise across firms but especially by high markups at the top 10% of firms. This also indicates an increase in market power since 1980.
Higher market power is also suggested by the increase in payouts to shareholders (dividends and share buybacks) over time. Payouts were 2-3% of assets in the 1970s but reached 6% by 2016, indicating firms were generating more profits from market power to distribute to owners.
Though the metrics differ in the 1950s-60s, the evidence agrees market power has risen since around 1980, based on a rise in economic profits, markups, and payouts to shareholders over this period.
Gutiérrez and Philippon found that business investment in new capital fell over the past few decades, even as payments to shareholders increased. This suggests that much of the payouts to shareholders were economic profits rather than payments for capital.
In economic theory, increased market power leads firms to restrict output and raise prices, generating more economic profit but reducing production. However, if market power rises across many firms, the declines in labor and capital demand can lead to lower input costs, which may not reduce output much.
Baqaee and Farhi find that increased aggregate markups are due to a shift in spending toward high-markup firms, especially in services. This reallocation raised productivity growth, so increased market power did not cause the slowdown.
But increased market power was associated with less firm investment in physical capital and R&D. This is consistent with theory - firms with market power restrict output and investment to keep prices and profits high.
So while market power did not reduce economy-wide productivity growth, it may have contributed to weaker business investment. There is some ambiguity about its role in the growth slowdown. Reducing market power could still raise productivity toward its maximum potential.

Here are the key points from the summarized passage:

Figure 10.1 plots the net investment rate (investment minus depreciation, as a share of operating surplus) for different business types from 1950-2016. It shows a downward shift for all business types starting around 2000.
This decline in investment rate is puzzling given that the Q ratio (a measure of expected firm performance) was high in the 2000s compared to historical levels (Figure 10.2). High Q ratios typically predict high investment.
One explanation is increased concentration and market power of firms. Figure 10.3 shows employment shifting toward larger firms, consistent with increased concentration.
Analyses by Gutiérrez and Philippon find that industries with the largest increases in concentration had the biggest declines in net investment, even controlling for other factors.
The implication is that increased industry concentration and market power may be causing the investment slowdown, not expectations of weaker growth. Around 10% of the capital stock is “missing” due to lower investment compared to what the high Q ratio would predict.
Market power can arise from limits on supply (e.g. consolidation leading to fewer firms) or from demand (e.g. consumers willing to pay a premium for a valued product).
Market power creates incentives for investment and innovation. Without some market power, there would be little economic growth.
The goal should not be eliminating market power, but finding the optimal level that provides incentives while avoiding excessive monopoly power.
Paul Romer’s research showed how innovation drives growth in modern economies. Unlike physical goods, ideas can be used by multiple people simultaneously (they are “nonrival”).
For people to invest time creating new ideas, they need to be rewarded somehow. Patents and other intellectual property rights create temporary monopolies, allowing innovators to profit from their ideas.
This market power incentivizes innovation, but too much monopoly power can also slow growth. The goal is to balance incentives for innovation with diffusion of new ideas.
Romer’s insights help explain the role of market power in growth. Some degree of market power is necessary for innovation, but excessive power can impede growth. The key is finding the right balance.
Paul Romer highlighted the difference between rival inputs like physical capital and nonrival inputs like ideas and technologies. Rival inputs can only be used in one place at a time, while nonrival inputs can be used simultaneously by many.
Nonrival inputs like ideas are key to economic growth because they can spread and raise productivity without diminishing. Rival inputs like capital are limited in their ability to raise total output.
For people to invest in creating new nonrival ideas, they need to be able to exclude others and charge a markup. This provides the incentive to innovate.
Some market power is necessary to incentivize innovation through markups, but too much power reduces competition and the pressure to innovate.
Evidence shows innovation peaks at an intermediate level of market power - enough to profit from new ideas but not so much that competition is stifled.
Recent rises in market power have correlated with less investment and innovation, suggesting we are on the wrong side of the “innovation hill” where too little competition reduces the incentive to innovate.

The implication of increasing market power and markups for firms is that it may have gone past the “sweet spot” where competition induces innovation and investment. While firms can charge high markups, they may not face enough competition to innovate at an optimal rate.

Reining in market power could potentially increase productivity growth. However, determining the optimal level of market power is difficult on a case-by-case basis. For some firms, high markups represent desirable products, while for others it may indicate restricted supply and collusion.

Intellectual property rights (IPRs) like patents and copyrights are one area where market power expansion may have gone too far. IPRs have expanded significantly in recent decades, allowing things like simple software and business models to be patented, with questionable benefits for innovation.

Antitrust enforcement is meant to challenge excessive market power, but cases have declined significantly since 1970, even as the number of firms has grown. This declining antitrust action coincides with rising market power, though the exact relationship is unclear.

Overall, while some firms likely have excessive market power, determining the optimal level is complex. A case-by-case approach is needed to balance affordable access to current ideas with incentives for future innovation. Trends like strengthened IPRs and weakened antitrust suggest market power may have expanded past an optimal point for the economy overall.

Here is a summary of the key points about reallocations across firms and jobs:

Productivity can increase when labor and capital are reallocated from less productive uses to more productive uses. This reallocation happens when businesses close down or expand, when workers switch jobs or get reassigned, and when capital gets reused by new owners.
The rate of reallocation between establishments has fallen in recent decades. Since reallocation is an important driver of productivity growth, this decline likely contributed to slower productivity growth.
Detailed data shows that for some manufacturing industries, about 40% of productivity growth comes from net entry of new establishments. As inputs shift into new, more productive establishments, overall productivity increases.
Movements of inputs between existing establishments contribute little to productivity growth. The main driver is resources shifting from exiting low-productivity businesses to new higher-productivity ones.
Similar patterns hold in the service sector. Reallocation across establishments accounts for a substantial share of productivity growth in industries like retail trade and banking.
Slower reallocation, due to less creation of high-productivity establishments and less exit of low-productivity ones, is a plausible source of the productivity slowdown. But the precise contribution is difficult to quantify.
Productivity growth in retail between 1987-1997 was largely due to new, more productive retail establishments replacing old, less productive ones rather than existing establishments becoming more productive.
This “shuffling” of inputs between establishments boosted retail productivity growth, even though technological change was minimal. It shows productivity growth is not just about technology but better organization and management.
The rates of establishment entry and exit, representing this “shuffling”, have declined over time. In the 1970s around 17% of establishments were new entrants and 13% exits per year. By 2000 it was 12% and 11% respectively.
In absolute terms, there is still a large turnover of establishments each year, with hundreds of thousands entering and exiting. But net establishment growth has slowed.
Slower turnover likely contributes to slower productivity growth, as inputs shift from low to high productivity establishments less frequently.
A similar pattern is seen in job turnover - job creation and destruction rates have declined over time. This further slows the reallocation of inputs to more productive uses.
The evidence suggests slower firm/establishment turnover has been a drag on productivity growth, though the exact magnitude is unclear.
Job creation and destruction capture labor market turnover, not just firms opening and closing. This matters for productivity as workers can move from low- to high-productivity jobs.
Job creation and destruction rates have trended downward over time, from around 22% and 15% respectively in 1976 to 14% and 12% in 2014. This slowdown reduces opportunities for productivity-enhancing job changes.
In absolute terms, around 15-16 million jobs are created and 14 million are destroyed per year. Job creation exceeds destruction, leading to net job growth over time.
The total number of jobs rose from 65 million in 1976 to around 120 million before the Great Recession, though it has yet to recover to the 2008 peak.
The slowdown in job reallocation likely contributed 0.1-0.15 percentage points to the productivity growth slowdown. This is a meaningful amount, similar to the effect of shifting from goods to services.
Potential reasons for the slowdown include increased market power dampening firms’ responses to shocks and slower population growth limiting labor force expansion. The causes are still debated.

Here is a summary of the key points about the decline in geographic mobility:

The number of people moving within the U.S. each year increased from the 1950s to the early 1980s, reaching over 45 million, but has declined since then to around 35 million in 2016.
As a percentage of the population, the share moving dropped from 20% in the 1950s to 11% in 2016.
There was less migration between states, metropolitan areas, and counties starting in the 1980s. For example, the share moving between states fell from 2.75% to 2% between 1980 and 2010.
This decline in mobility matters because productivity, measured by GDP/GSP per worker, varies widely across locations. Some states and cities are over 2 times more productive than others.
Certain places like Silicon Valley have much higher productivity, so less mobility reduces the ability to reallocate workers to more productive places.
Reasons for declining mobility could include increasing housing costs in productive cities and changing preferences of workers.
Regardless of the cause, reduced geographic mobility within the U.S. is a potential source of slower productivity growth in recent decades. Matching workers to the most productive locations is more difficult with low mobility.
There is a positive relationship between the relative GDP per worker in a metropolitan area and its size - larger cities tend to be more productive. However, people have not been moving from low-productivity areas to high-productivity areas as would be expected.
One reason is that people have been moving south and west, toward warmer climates, rather than toward the most productive cities. This migration pattern drags on productivity growth.
Another reason is constraints on housing supply in productive cities like Seattle. High demand raises housing prices, which limits in-migration. This is similar to the effect of market power limiting competition.
Housing can act like a profitable industry with market power when increased demand translates into higher prices rather than increased supply. We can see evidence of this by looking at economic profits of housing as a share of housing value-added.

Here are the key points about whether the government caused the slowdown in economic growth:

Tax rates and regulation are often blamed for slowing growth, but the evidence does not clearly support this. Total tax revenue as a share of GDP has been roughly constant over recent decades. Regulatory burden may have increased some but does not appear to explain the timing of the growth slowdown.
Inequality rose starting in the 1980s, well before the slowdown in growth. There are theoretical reasons why inequality could lower growth, but empirically this relationship is murky. The rise of inequality does not line up well with the growth slowdown.
Trade with China likely did reduce some manufacturing jobs and wages, but the aggregate effect on growth appears small. Only about one-fifth of the slowdown can plausibly be attributed to import competition from China.
Overall, changes in government policy, inequality, and trade do not appear to explain much, if any, of the significant slowdown in productivity and output growth that occurred in the 2000s. The evidence better supports the importance of slower human capital growth, the shift to services, and possibly decreased dynamism.
There is little empirical evidence that changes in taxes or regulations have had a major impact on U.S. economic growth rates.
The Bush tax cuts of 2001 and 2003 did not lead to higher growth rates. Research on the 2003 dividend tax cut found no effect on business investment.
Kansas implemented large tax cuts in 2011-2012 but did not see an increase in growth rates compared to the rest of the U.S. There was no “jump start” of the Kansas economy.
Studies find little effect of individual income tax rates on labor supply and participation. There was no large increase in tax rates in the early 2000s that could explain the growth slowdown.
One study looked at industry-level regulation data and did not find a strong link between increased regulation and declines in startups, job creation, and job destruction.
Overall, changes in taxes and regulation do not seem able to account for a sizable share of the decline in U.S. growth rates starting in the early 2000s. Other factors must be more important for explaining the growth slowdown.
Research finds no clear relationship between industry regulation and firms starting up or job turnover. In fact, more regulation is sometimes associated with more job creation.
There is no clear link between a state’s ranking on the ALEC-Laffer index of low taxes/regulation and its GDP per worker level or growth. Highly ranked states don’t consistently outperform.
Housing regulations are tightest in highly productive metro areas like San Francisco, New York, Seattle. This limits mobility into productive cities, slowing growth. The extent regulations increased over time is unclear.
Overall, aside from housing regulations, there is little evidence taxes and regulations significantly affected the slowdown in economic growth since 1990. The reasons for the lack of effect are unclear.

Here is a summary of the key points regarding whether inequality caused the economic growth slowdown:

Rising inequality coincided with the growth slowdown, leading some to wonder if inequality was a cause of the slowdown. However, inequality is better viewed as another symptom of increasing corporate market power rather than a separate cause.
Inequality stems from the same root cause as the slowdown - increasing monopoly power of firms. As firms gain more market power, they are able to charge higher markups and extract more profits. This transfers income from workers and consumers to firm owners and shareholders, increasing inequality.
There is no evidence that inequality on its own reduces incentives to work and invest. Extreme inequality could potentially cause political instability and unrest that damages growth, but the level of inequality in developed countries has not reached that point.
Inequality may reduce opportunities for poorer children, decreasing intergenerational mobility. However, this effect likely operates over longer time horizons and cannot explain the abrupt slowdown around 2000.
Overall, while rising inequality is concerning for many reasons, it does not appear to be a primary cause of the growth slowdown. The slowdown is better explained by increasing corporate market power and weaker competition. Policies to address inequality should focus on containing market power and expanding opportunities.
Income inequality increased substantially from the 1960s to the 2010s, with the top 1% of earners seeing their share of national income double from around 12.5% to 20%.
Much of this increase came from a rise in labor income (wages, salaries, stock options) going to top executives and financial professionals rather than dividends or capital gains. These “supermanagers” account for a large fraction of the top 0.1% of earners.
It is unclear whether rising inequality directly contributed to the economic growth slowdown. Stagnant incomes may have reduced investment in education and human capital growth. Greater income concentration may have shifted demand toward services. But these effects do not seem large enough to explain much of the slowdown.
There is no definitive link between inequality and growth rates. Inequality may impact components like physical capital accumulation or productivity, but the relationship is complex. Overall, rising inequality alone does not directly explain the growth slowdown.
The rise in inequality itself does not seem to be a major cause of the growth slowdown, based on its limited effects on human capital acquisition and the shift toward service industries.
Inequality exacerbated the shift out of goods production, but only to a minor degree - the general rise in living standards is more important.
The effects of rising inequality on human capital and the shift into services appear marginal and cannot explain much of the growth slowdown.
However, the rise in inequality does corroborate the story of increased market power of firms, as more income flowed to executives.
Regarding trade and China, imports do not mechanically lower growth or GDP levels. The “imports lower growth” argument is a misinterpretation.
However, research links Chinese imports to declines in US manufacturing employment. This could have contributed somewhat to the slowdown by reducing human capital or accelerating the shift toward lower productivity growth industries.
But the size of the effects of trade and China specifically look too small to account for much of the growth slowdown. China and trade are minor factors compared to the other larger issues discussed.
The statement “imports reduce GDP” is incorrect. This mistake comes from misinterpreting the national income accounting identity.
Increasing imports does not mechanically reduce GDP. The accounting identity just shows how GDP is allocated to different spending categories.
Adding imports to both sides of the equation clarifies that imports add to the total amount of goods and services available.
Imports competing with domestic industries can reduce employment in those industries. But displaced workers often have trouble transitioning to new jobs.
So while imports don’t directly reduce GDP, adjustment frictions when workers change industries can temporarily reduce production and growth. The statement is still wrong, but trade can have real negative effects.
The degree to which different commuting zones (CZs) were exposed to competition from Chinese imports depended on the industries located there. For example, furniture CZs in North Carolina and Tennessee faced a lot of competition, while auto CZs in Alabama and South Carolina did not.
Across all CZs, for every $1,000 increase in imports per worker, the unemployment rate rose 0.2 percentage points and the share of working-age people not in the labor force rose 0.5 percentage points.
The most exposed CZs saw imports per worker rise by about $4,300. This increased the unemployment rate by 1 percentage point and the share out of the labor force by 2.4 percentage points.
Increased trade did not cause people to leave affected CZs - they stayed but became unemployed or left the labor force. The reallocation effects that would mitigate trade costs did not occur.
Between 1990-2000, trade lowered manufacturing employment by 548,000. Between 2000-2007, it lowered it by another 982,000, for a total of 1.53 million jobs lost.
This accounted for 21% of the 1990-2007 decline in manufacturing employment. It lowered labor force participation by 1 percentage point and raised unemployment by 0.37 percentage points.
The effect of trade with China on growth rates was small - it may have lowered productivity growth by around 0.016% per year. Overall, trade with China contributed little to the growth slowdown.
The accounting in Table 17.1 likely understates the influence of “success” (i.e. higher living standards, smaller families, etc.) on the growth slowdown and overstates the influence of possible “failures.”
Reversing the demographic shift and aging population that contributed to slower growth would require sacrificing advances in living standards and rights that few would support.
Boosting growth by shifting production back to goods would require destroying much of our existing capital stock, as happened in WWII, which few would support.
China’s high past growth was partly due to its lower living standards relative to the US. Few would accept reducing living standards to US levels of decades past just to increase growth.
The growth slowdown is largely an unintended consequence of choices that improved living standards. Success, not perfection.
Problems like increased inequality and market power are still valid issues to address through policy, even if their impact on growth is unclear. The focus should be on distributional impacts, not growth effects.
Similarly for trade - policies should focus on compensating losers, not boosting growth.
The growth rate impact of changes in spending, regulation, and taxes is likely small. Focus should be on who benefits and who is hurt, not aggregate growth effects.
There is likely no way to reverse the overall growth slowdown through policy changes. Attempts to boost growth via changes to market power, trade, taxes, regulation, or housing would have limited effects.
The one exception is immigration policy. Increasing immigration, particularly of skilled workers, could offset some of the growth slowdown caused by declining population growth and aging. Adding 255,000 working-age immigrants per year could reverse the 0.35 percentage point drag from declining worker-to-population ratio.
Growth is expected to stay low or decline further in coming decades due to ongoing demographic changes, shifts to services, and reductions in work hours. Productivity growth in services may improve but is unlikely to accelerate growth substantially.
New technologies like AI and gene editing have potential to transform lifestyles and production, but unclear if they will boost GDP growth rate. Their risks may outweigh potential small growth gains.
The growth rate alone does not determine economic well-being or progress. We should evaluate growth effects along with distributional impacts. Our choices drive growth, and slower growth may reflect positive lifestyle changes.

Here are some key details on the data sources and calculations for human capital in Chapter 3:

Data on total employees age 16+ is from the Bureau of Labor Statistics (via FRED database).
Data on total population is from the Census Bureau (via FRED).
Data on educational attainment is from the Census Bureau’s Current Population Survey (CPS).
Returns to education are based on estimates from Card (1999) and Lemieux (2006):
- Return for high school diploma: 15%
- Return for some college: 5%
- Return for bachelor’s degree: 45%
- Return for advanced degree: 60%
Returns to experience are also based on Lemieux (2006):
- Return for 1st year of experience: 8%
- Return depreciates by 0.5% per year after that

To calculate human capital:

Take the number of people in each education bracket from CPS data
Multiply by the return for that education level
Take number of employees in each experience bracket
Multiply by the return for that experience level
Aggregate across education and experience brackets
Divide by total population to get human capital per capita

So in summary, human capital combines data on educational attainment, returns to education, workforce experience, and returns to experience. The sources are all standard US government data, and the returns are based on empirical labor economics estimates.

Here are the key sources and details for the growth accounting in Chapter 4:

The idea of calculating “residual” growth originated with Tinbergen (1942) and became widespread after Solow (1957). Issues in calculating productivity growth are discussed in Stiroh and Jorgenson (2000) and Jorgenson, Gollop, and Fraumeni (1987).
Solow used labor’s share of total output, which can be problematic unless strict conditions hold (Hall 1988, 1989).
Accounting for the fact that capital is produced using GDP is from Klenow and Rodriguez-Clare (1997) and Hall and Jones (1999).
Accounting for real GDP per capita differs from accounting for real GDP due to changes in hours worked per person. This issue is discussed in Fernald (2014).
The growth accounting is done by calculating the contribution of changes in inputs and total factor productivity to GDP growth. The main inputs are labor (hours worked) and capital.
Labor input is calculated using hours worked from the BLS and employment from the Census. Capital input uses data on investment and capital stocks from the BEA.
The labor and capital shares are calculated as the total employee compensation and gross operating surplus shares of GDP from the BEA National Income Accounts.
Total factor productivity is calculated as a residual, representing growth not accounted for by changes in measured inputs.
The calculations show the slowdown in GDP growth since 2000 was driven by slower growth in capital deepening and especially total factor productivity. Changes in labor input were less important.
Additional details on data sources and methods are provided in the human capital and physical capital sections. But the overall approach follows standard growth accounting techniques.
The elasticity of real GDP with respect to physical capital can be inferred from the share of physical capital in total input costs, under an assumption of cost minimization and constant returns to scale.
The formula for accounting for growth in real GDP per capita in terms of physical capital, human capital, and residual growth is provided. The residual is calculated as the difference between actual growth and the contributions of the two capital inputs.
It is argued that growth in physical capital depends in part on growth in real GDP per capita, so some adjustments are needed to avoid double counting this effect. A formula is provided to properly account for the contribution of physical capital.
Similar accounting can be done using real GDP per worker rather than per capita. The contributions of physical and human capital are found to be somewhat smaller in this case.
Different assumptions about measuring human capital can lead to changes in the accounting. An alternative human capital series from Fernald and Jones (2014) is used to illustrate this.
There are uncertainties in this type of growth accounting, so precision in the numbers should not be overinterpreted. The general patterns and magnitudes are more important.

Here are the key points from the appendix notes related to chapters 5-7:

Chapter 5 - The Effect of an Aging Population

Data on age structure, dependency ratios, and fertility rates are from OECD and Our World in Data. Figures in the chapter are based directly on this raw data.
Table A.1 shows accounting for growth in real GDP per worker over time periods from 1950-2000, broken down into contributions from physical capital, human capital, and residual growth.
Evidence suggests aging population and recent recession both contributed to decline in labor force participation, but aging appears to be a bigger factor based on state-level data.
Theory and evidence relate higher incomes to lower fertility over time within the US. Across countries, long-run growth is associated with lower and delayed fertility.
Access to birth control pill causally reduced early births and increased women’s workforce participation and hours based on policy variation across states.

Chapter 6 - The Difference between Productivity and Technology

Recent work by Gordon and others suggests slower technological change contributed to growth slowdown, though connection to productivity growth is not always clear.
Measurement issues make it hard to capture new technologies, so productivity underestimates true technological change.
Evidence suggests new technologies are harder to find over time.
Growth in research workers has slowed since 1970s.

Chapter 7 - The Reallocation from Goods to Services

Industry-level data from EU KLEMS used to analyze sectoral shifts and contributions to productivity growth.
Slowing sectoral reallocation of labor from low to high productivity industries contributed to productivity slowdown after 1970.
Rising health care spending not mainly due to aging population; technology and institutions play roles.
Baumol’s cost disease driven by lack of productivity growth in some services, though its quantitative importance is debated.
The BEA industry classification system divides the economy into 17 major industries, labeled A-Q.
Codes M-N cover professional, scientific, technical, administrative, and support services.
Codes R-S cover arts, entertainment, recreation, and other services.
Code O-U covers community and personal services, but this is broken down further into:
Code O for public administration
Code P for education
Code Q for health and social work
To calculate aggregate productivity growth, take the productivity growth rate for each industry multiplied by its value-added share of GDP.
The shift toward services is driven by both relative price changes and preferences. As goods become relatively cheaper due to higher productivity growth, expenditure shifts toward services even if their prices are rising. This is because goods and services are complements - people are reluctant to substitute between them.
The cost disease theory explains how services can become more expensive but also absorb more resources: productivity grows faster in goods, driving down relative prices, so expenditure shifts to slowly-growing services.

Here are some key points about market power, productivity, and investment:

Economic profits have increased significantly since the 1980s, suggesting rising market power and markups by firms. This is seen across advanced economies.
Higher markups can raise measured productivity growth if they reflect a shift of resources towards high markup industries. The total value produced rises.
However, rising market power can reduce investment incentives for firms, since they rely less on expanding market share to earn profits. This may partly explain the decline in capital investment since the 2000s.
Some degree of market power and profits may be necessary for innovation and growth. Schumpeterian theories argue temporary monopolies incentivize innovation that ultimately benefits consumers. The relationship between competition and innovation may be U-shaped.
Overall, the evidence suggests increasing market power has likely played a role in shaping macroeconomic trends in productivity, investment, and innovation. Policymakers face a tradeoff between promoting competition while still providing incentives for innovation and growth.

Here are some key points about the evidence presented in this section on whether government policies caused the economic slowdown:

The 2003 dividend tax cut led to increased dividend payouts but did not appear to substantially increase investment or employment (Yagan 2015, Chetty and Saez 2005).
Higher state tax rates are associated with less innovation and entrepreneurship (Akcigit et al. 2018).
The elasticity of taxable income with respect to tax rates appears to be modest based on changes in federal income tax rates (Saez, Slemrod, and Giertz 2012).
There is mixed evidence on whether reported taxable income of high earners responds to tax rate changes. The 1986 and 1993 federal tax changes showed some response, but the 2013 increase did not (Saez 2004, 2016).
Regulatory restrictions increased over time, and more regulation is associated with less business dynamism (Al-Ubaydli and McLaughlin 2015, Goldschlag and Tabarrok 2018).
Overall, changes in taxes and regulation likely played some role in the economic slowdown but do not appear to be the primary driver based on the empirical evidence. Factors like market concentration and intangible capital seem more central to explaining declining dynamism and growth.

Here is a summary of the key points about the data sources in Chapter 14:

Data on state-level GDP per worker and metropolitan area (MSA) GDP per worker comes from the same sources as Chapter 13.
The American Legislative Exchange Council (ALEC) index of state business conditions is from the ALEC website.
Housing regulation data originally comes from Gyourko, Saiz, and Summers (2008). The author obtained this data from their website.
World Bank Doing Business indicators are from the Doing Business website.
There is no spreadsheet source for the ALEC index. The author entered the values manually into a text file.
The housing regulation data uses different MSA codes than current codes, so the author built a crosswalk using data from the NBER website to match up the MSAs.
The code and crosswalk used to match up MSAs are available on the book’s website.

In summary, the data comes from a variety of sources including ALEC, the World Bank, and academic papers. The author collected the data from spreadsheets and websites, and in some cases manually entered or matched up the data. The code and crosswalks are available on the book’s website.

Here is a summary of the key points from the references:

Hion and Howitt (2009) provide an overview of economic growth theory, including models of technological change, human capital accumulation, and institutional development.
Akcigit et al. (2018) analyze the impact of taxation on innovation in the 20th century, finding that corporate and estate taxes had negative effects while personal income taxes had little effect.
Al-Ubaydli and McLaughlin (2015) introduce a new database on industry-specific regulations that enables quantitative analysis of the impacts of regulation.
Autor and co-authors have written seminal papers on job polarization, the China trade shock, and the growth of low-skill service sector jobs.
Bailey (2006, 2010) analyzes the impacts of access to oral contraceptives on women’s career and family choices.
Bakija et al. (2008) examine the causes of rising top income inequality, pointing to executives and finance professionals.
Banerjee and Duflo (2003) survey the empirical literature on inequality and growth.
Bloom et al. (2017), Gordon (2016, 2018), and Cowen (2011) discuss the potential slowing of innovation and productivity growth.
Boldrin and Levine (2002, 2008, 2013) make the case against intellectual property protections.
Card (1999) and Card and Peri (2016) study the labor market impacts of education and immigration.
Fernald, Hall, Stock, and Watson (2017) examine the slow recovery of output after the Great Recession.
Goldin and Katz (2007, 2008) analyze the race between education and technology in 20th century economic growth.
Greenwood, Seshadri, and others study family formation and fertility choices.
Gutiérrez and Philippon (2017) find a decline in business investment despite rising corporate profits.

Here is a summary of the key points from the passages:

Real GDP per capita growth in the U.S. slowed from over 2% per year in the decades after World War II to around 1.5% more recently. Understanding the causes of this slowdown is important for policy.
Growth accounting shows the slowdown is mainly due to slower growth in total factor productivity, not slower capital accumulation. This points to problems with innovation and dynamism in the economy.
Possible explanations include mismeasurement of outputs and inputs, especially in ICT industries and the growing service sector; slower idea production; market power and rent-seeking by dominant firms; demographic shifts; and increased regulation.
Determining the relative importance of these factors is challenging but critical. Focus areas include measuring intangible capital, assessing concentration and competition, evaluating impact of aging workforce, and examining growth at the regional level.
Policies to promote innovation, reduce barriers to reallocation, increase openness, reform outdated regulations, improve education and training, welcome high-skilled immigration, and reduce inequality may help boost growth. But more research is needed to provide robust policy guidance.

Here are the key points I gathered from summarizing the provided index:

Productivity growth has slowed in recent decades despite technological progress. Differences exist between productivity and technology.
Labor force participation and human capital accumulation are important drivers of growth historically.
Physical capital investment affects productivity. Market power can reduce investment incentives.
Regulation and trade influence productivity through impacts on reallocation and competition.
The shift from goods to services has affected measured productivity growth. How to accurately measure services output is debated.
Market power has increased in many industries, possibly dampening growth. Causes and policy implications are disputed.
Inequality has risen with slowing productivity growth but the relationship is complex. Theories exist on both directions of influence.
Understanding local and sectoral differences can provide insights into the productivity slowdown. Metro areas and specific industries show varying trends.
Multiple factors are likely involved in declining growth, including demographics, capital investment, regulation, trade, and market power. Their interactions and measurement are active research areas.

#book-summary

Fully Grown Why a Stagnant Economy Is a Sign of Success - Dietrich Vollrath