Macro Roundup

Ufuk Akcigit, Jeremy Pearce and Marta Prato, “Tapping Into Talent: Coupling Education and Innovation Policies for Economic Growth,” September, 2020, https://www.nber.org/papers/w27862

Time lag of investment in human capital versus R&D subsidies“…Fourth, we solve for the transitional dynamics of our model and find that it takes time for all of these policies to show their full effect. We find that R&D policy is the most effective policy for innovation in the short-run. On impact, R&D policy stimulates the purchase of R&D inputs other than human capital (e.g. lab equipment) by the existing researchers, which translates into more innovation. Education policy, on the other hand, has limited effectiveness in the short-run due to time-to-build human capital, but is more effective in the medium to long-run. It takes around five years for educational subsidies to surpass R&D policy in terms of its overall growth effect for equivalent expenditure of 0.5% of GDP….”

Human capital most important factor“….Third, our analysis suggests that education policy is more effective in more unequal societies in stimulating innovation. We analyze the responsiveness of policy under different levels of financial frictions. In an environment with lower financial frictions, fewer individuals are unable to pay for schooling; thus, subsidizing the cost of education is less effective in stimulating economic growth. In an economy where no one is constrained, educational subsidies have no effectiveness, as individuals who want to pursue a career in research will do so regardless. In an economy with full frictions, educational subsidies are the most effective policy tool for innovation….”

Here they look at the utility of the policy toolkit“…Second, there is a pecking order between education and R&D policies. The optimal policy mix depends on the size of the budget the government has to allocate to research. If the government has 0.5% of GDP budget to allocate to research, our framework predicts it should allocate all of its resources towards educational subsidies to improve the talent pool by enabling access to education for financially constrained individuals. With a larger budget, the government should mix subsidies to education, subsidies to R&D, and an expansion in the supply of education slots. Given a budget of 1% of GDP, the optimal allocation is between education and R&D subsidies, with 50% of the budget for R&D subsidies and 50% to educational subsidies. Given a budget of 2.5% of GDP, it is optimal to allocate 58% of the budget to R&D subsidies, 26% to education subsidies, and 16% to expanding the educational slots available. The fact that there is an optimal mix highlights the role of the complementarities of the policies in their contribution to economic growth…”

They have four main findings in terms of human capital’s impact on innovation, “…First, we show that, in steady state, R&D policy has limited effectiveness compared to benchmark models of endogenous growth. R&D policy has even more limited contemporaneous effects. A 10% R&D subsidy only has a contemporaneous effect on the growth rate of about 0.6%, because of the time-to-build component of human capital. Further, the long-run effects of R&D are smaller than standard models due to the fact that the talent pulled into research is restricted by both educational capacity and financial constraints. A 10% R&D subsidy has a long-run effect on the growth rate of 5%, compared to 10% in standard models…”

“…Starting in 2002, the Danish government required universities to double PhD enrollment in the span of 10 years, with the goal of expanding the pool of scientists and inventors. As highlighted in Fact 9, the increase in PhD enrollment was associated with a decline in average IQ of enrolled PhDs. This empirical finding illustrates constraints the government faces when attempting to expand the pool of researchers and inventors due to the scarcity and heterogeneity of talent, which is a key determinant of innovation outcomes. It also stresses the importance of tapping into other sources of talent in the economy: attracting into research those talented individuals who are financially constrained or work in other sectors….“…Figure 14 shows that as the number of slots for PhDs increases, the average IQ of the enrolling students falls. This indicates that there is heterogeneous quality of enrollees and expanding slots may draw in a marginal researcher less talented than the average researcher from the existing pool. Thus, even though policy can increase the supply of researchers, there is a trade-off between expanding the pool of PhDs and the average talent of PhDs in the economy….In order to speak to innovation and the effects of policies, we match the dataset to external patent data from the European Patent Office (EPO) and policy data from the Danish Ministry of Education. A unique feature of this dataset is detailed information on an innovation and education policy change implemented in 2002, which introduced new R&D subsidies and dramatically increased funding to universities and the level of PhD enrollment. We exploit this variation to better isolate the links between talent, education and innovation. We stress ten facts in our framework made possible by the extensive Danish dataset that shape our analysis:Fact 1 Individuals with higher IQ are more likely to obtain a PhD.Fact 2 Individuals with higher parental income are more likely to obtain a PhD. Fact 3 Individuals’ IQ is correlated with parental income, but not perfectly.Fact 4 Only a fraction of people with high IQ and high parental income obtain a PhD. Fact 5 PhDs are 20-times more likely to become inventors compared to the average person in society.Fact 6 Conditional on education level, higher IQ people are more likely to innovate.Fact 7 Inventors work in teams and team size is heterogeneous.Fact 8 The probability of innovating as a team leader over an inventor’s life-cycle follows an inverted-U shape. Fact 9 An increase in the number of PhD slots is associated with a decline in the average IQ of PhDs.Fact 10 The economy is open in the goods market but closed in the skills market: import and export accounted for 44% and 51% of GDP in 2010, but foreign migrants make up only 10% of inventors in Denmark…”

New Akcigit using a unique natural experiment/Danish dataset looks at the links btw higher education policy, R&D subsidies and innovation. His data suggests that talent is the limiting constraint.

David Cutler and Lawrence Summers, “The COVID-19 Pandemic and the $16 Trillion Virus,” Journal of the American Medical Association, October 12, 2020, https://jamanetwork.com/journals/jama/fullarticle/2771764

“…. The Congressional Budget Office projects a total of $7.6 trillion in lost output during the next decade…. In addition toCOVID-19deaths, studies suggest increased deaths from other causes, amounting to almost 40% of COVID-19-related deaths. Thus, if the current trajectories continue, an estimated 625 000 cumulative deaths associated with the pandemic will occur through next year in the US…. In environmental and health policy,3 for example, a statistical life is assumed to be worth $10 million. With a more conservative value of $7 million per life, the economic cost of premature deaths expected through the next year is estimated at $4.4 trillion… Data from survivors ofCOVID-19suggest that long-term impairment ccurs for approximately one-third of survivors with severe or critical disease.4Because there are approximately 7 times as many survivors from severe or critical COVID-19 disease as there are COVID-19 deaths, long term impairment might affect more than twice as many people as the number of people who die… an estimated loss from long-term complications of $2.6 trillion for cases forecast through the next year… The proportion of US adults who report symptoms of depression or anxiety has averaged approximately 40% since April 2020; the comparable figure in early 2019 was 11.0%.5 These data translate to an estimated 80 million additional individuals with these mental health conditions related to COVID-19…. the valuation of these losses could reach approximately $1.6 trillion….. related to the lost output and health reduction are shown in the Table. The total cost is estimated at more than $16 trillion, or approximately 90% of the annual gross domestic product of the US. For a family of 4, the estimated loss would be nearly $200 000. Approximately half of this amount is the lost income from the COVID-19-induced recession; the remainder is the economic effects of shorter and less healthy life

The residual are long term health/mental health impact of the pandemic.

The largest component of that is lost GDP,he’s using CBO estimates at $7.6T over the next decade. The second largest component is deaths (they value a life a $7mm) and are estimating 625,000 excess deaths for $4.4 Trillion (though they are estimating 450,000 directly from Covid)

Just an FYI, Summer’s tacks a whack at estimating Covid’s economic impact. Bottom line is $16 Trillion (90% of a years GDP)

Core of their employment argument, “…. how much the stock market reflects the economy is an empirical question. Employment is a reliable benchmark for the state of the economy.Employment data for the whole economy is available for our sample period, does not depend on accounting rules, and is collected by the Federal government irrespective of whether a firm is public or not. For public corporations, employment is available for most corporations annually since the early 1970s. We find that, from 1973 to 2019, the percentage of employees working for public firms is highest at the start of the sample period and falls sharply in the 1980s. At the beginning of that period, more than 41.4 percent of non-farm workers in the private sector work for public firms, but in 2019, that percentage is 29.0 percent. In 2019, the percentage of non-farm workers in the private sector working for public firms is the lowest for our sample period for industrial firms and for all firms, including financial firms, it is the lowest except for 1989 and 1990. We create a measure of unrepresentativeness to capture how poorly market capitalization proxies for a firm’s contribution to employment over time. For each firm, we measure the absolute value of the difference between the firm’s market capitalization as a share of the market capitalization of the stock market and the firm’s employment share among public firms defined as the firm’s employment divided by the employment of all stock market firms. The measure of unrepresentativeness is proportional to the sum of these differences. Intuitively, it is the difference between a portfolio that holds the stock market and a portfolio with weights equal to the employment weights of firms. The unrepresentativeness measure increases as the two portfolios differ more. We call this measure our employment unrepresentativeness measure. This measure follows a w-shape, with its lowest values in the 1980s and 1990s and its highest values in the 1970s, around 2000, and in the recent past. The stock market is more unrepresentative at the end of our sample period with respect to employment than at any time except around the year 2000. Another way to show that market capitalizations are not instructive about firms’ contribution to employment is to estimate how much of the variation in market capitalizations can be explained by variation in employment. We find that from 1973 to 2019 a firm’s employment never explains as little of its market capitalization as in 2019. In most years in the 1970s and early 1980s, firm employment explains more than 50% of the variation in market capitalization. In each of the last four sample years but one, firm employment explains less than 20% of the variation in market capitalization…”

Core of their value added argument, “…Assessing a firm’s contribution to GDP is more fraught with difficulties. A firm’s value added depends on the total compensation it pays to its employees. However, most firms do not disclose how much they pay to their employees separately…. we find that the contribution of public firms to GDP is higher in the 1970s than in the 2000s.This result is insensitive to the approximations we make. Using the same approach to measure unrepresentativeness as with employment, we construct a measure of unrepresentativeness for value added. The stock market’s unrepresentativeness for value added follows a w-shape with unrepresentativeness high early in the sample period, around 2000, and late in the sample period. In contrast to employment, the unrepresentativeness at the end of the sample period is less than at the beginning of the sample period….We find that variation in value-added best explains the variation in market capitalization in the late 1970s and early 1980s. While employment explains less of the variation in market capitalization in the 2010s than in the 1970s, the same is not the case for value added. The variation in value-added explains the variation in market capitalization less well at the end of the sample period than in a number of years, but it explains it better than in the period from 1988 (1991 for industrial firms) to 2003…”

Conclusions, “….we examine how listed firms contribute to the economy over time and how their market capitalization is related to their contribution to the economy. We find that stock market firms as a group contribute less to employment and to GDP than in the 1970s. Since 1973, industrial stock market firms never contribute as little to employment as they do in 2019 and all firms on the stock market contribute less than in 2019 only two years.Though the contribution to GDP of stock market firms is less in recent years than in the 1970s, the decline in contribution to GDP is less than the decline in contribution to employment. Our estimates for the contribution of listed firms to GDP come with some important caveats. Listed firms generally do not disclose their employment expenses, so that we have to approximate them using approaches developed in the recent literature. Further, though listed firms disclose employment, they do not generally disclose separately domestic and foreign employment. As a result, our estimates of employment and value added for listed firms are estimates for worldwide employment and worldwide value added of these firms. To the extent that foreign activities of firms increase over time, our estimates likely understate the decrease in the contribution of listed firms to the US economy. It is well-known that the relative importance of investments in intangible capital has increased over time. These investments are mostly expensed under GAAP. Though we can adjust firm income for investment in intangible capital, we cannot adjust employment for employees who work on projects that increase invested capital. If we adjust income for investment in intangible capital, our conclusions about the evolution of the contribution of value added of listed firms to GDP are unchanged. We measure how representative the market capitalization of firms is of their contribution to the economy. We find that our measure of unrepresentativeness follows a w shape for both employment and value added: the measure is high in the 1970s, around 2000, and at the end of the sample period.For labor, unrepresentativeness increases over time, but not for value added. We also find that the relation between firm market capitalization and firm employment collapses during our sample period. In the 1970s, employment explains 50.7% on average of the variation in market capitalization across firms. In the 2010s, it explains 21.8% on average.A firm’s value added explains much more its market capitalization than its employment. This is true throughout our sample period. Strikingly, there is no consistent degradation in the ability of value added to explain the variation in market capitalization across firms. In the 1970s, value added explains 68.2% of the variation in market capitalization in average. In the 2010s, it explains 67.7%.Perhaps not surprisingly, the top market capitalization firm evolves in the same way as the market as a whole. In the 1950s, the top market capitalization firm was the top employer in the country. In the 2010s, it is the 40th. However, the top market capitalization firm is in the top ten of firms by value added in the 2010s as in the 1970s. The evolution we document is largely the result of the decline of manufacturing in the US. Not all firms are equally suited to be public firms. Small firms and service firms are much less likely to be listed on stock exchanges. Over our sample period, employment by the manufacturing industry falls sharply in importance while employment by the service industries grows dramatically in importance. A large fraction of the employees of manufacturing firms work for public firms. A small fraction of employees of service firms work for public firms. As a result, the fraction of employees working for public firms falls….”

This was a very interesting factoid (see their Table 5 below), “…. Strikingly, from 1950 to 2019, only seven distinct firms have the top market capitalization in at least one year. Only two of these firms, General Motors and AT&T, ever have the largest number of employees in the US. The percentage of US non-farm payrolls represented by the largest market capitalization firm declines steadily over time, so that it is at its lowest in the 2000s. The percentage of US GDP represented by the largest market capitalization firm is also lower in the 2000s than before 1980. The contribution of the top market capitalization firm to the economy in 2019 is a fraction of GM’s contribution in 1953. …”

They find representativeness is worst when the market is most highly valued and worsens over time for employment, but not for value added. “…we measure the extent to which a firm’s market capitalization reflects its concurrent contribution to the economy. We show a strong downward trend in the extent to which a firm’s market capitalization reflects its concurrent contribution to employment. With value added, market capitalizations are least instructive about a firms’ contemporaneous contribution to GDP around 2000, but market capitalizations appear to be as instructive about firms’ contribution to GDP in the 1970s as in the 2010s…..In the 1970s, employment explains 50.7% on average of the variation in market capitalization across firms. In the 2010s, it explains 21.8% on average.A firm’s value added explains much more its market capitalization than its employment. This is true throughout our sample period. Strikingly, there is no consistent degradation in the ability of value added to explain the variation in market capitalization across firms. In the 1970s, value added explains 68.2% of the variation in market capitalization in average. In the 2010s, it explains 67.7%...”

New NBER argues that there is a declining correlation btw a firm's share of employment and its share of stock market value.

“…Our data, which range from 1980 to 2016, allow us to characterize responsiveness in a more recent period than most of the literature, and our event-study approach enables us to estimate effects over different time horizons and test for dynamic effects. We then build a simple framework to relate these state-level effects to a policy-relevant elasticity at the national level, which is the state-level realization response after removing migration effects and accounting for average state taxes and a minor aggregation adjustment term. We find that this policy-relevant elasticity of realizations with respect to capital gains tax rates over a ten year period is approximately -0.3 to -0.5 depending on the specification, and that the estimates are larger in absolute value in the short and medium run than in the long run….This paper uses an event study approach to estimate the effect of capital gains taxation on realizations at the state level, and then develops a framework for determining revenue-maximizing rates at the federal level. We find that the elasticity of revenues with respect to the tax rate over a ten-year period is -0.5 to -0.3,indicating that capital gains tax cuts do not pay for themselves, and that a 5 percentage point rate increase would yield $18 to $30 billion in annual federal tax revenue. Our long-run estimates yield revenue-maximizing capital gains tax rates of 38 to 47 percent….”

Their takeaway, “….We highlight three implications of these elasticity estimates. First, these estimates are well below an elasticity of one in absolute value, which indicates that capital gains tax cuts do not pay for themselves. We formallytest and reject the null of an elasticity of -1.0. Second, these estimates suggest thatraising capital gains tax rates by 5 percentage points (in the current regime with unlimited deferral and step-up-basis at death) would yield 18 to 30 billion in annual tax revenue, which is roughly twice the amount suggested based on recent estimates from the Joint Committee on Taxation (JCT), which according to Gravelle (2020) currently uses an elasticity of -0.7 to score proposals.1 Third, ourlong-run elasticity estimates correspond to point estimates for the revenue-maximizing capital gains tax rates of 38 to 47 percent….”

Current state of capital gains taxes, “…Table 1 provides summary statistics. On average, state capital gains tax rates are 4 percent, but they range from 0 to 15 percent in our sample. Figure 1 plots the maximum federal tax rate over time. The current maximum capital gains tax rate is 23.8 percent. Combining both tax rates and accounting for interactions and phase-outs results in an effective keep rate of 75 percent on average, indicating that a dollar of realized capital gains amounts to 75 cents after taxes….”

Worth noting from Owen Zidar (Capitalists in the Twenty-First Century) looks at the impact of changes in treatment of capital gains on the state level to estimate a the revenue-maximizing rates for federal level. They estimate a Federal “…revenue-maximizing capital gains tax rates of 38 to 47 percent…”