This blog post is available as a PDF here

Summary­­

  • Economic growth is a key driver of long-term stock and bond returns.
  • Economic growth comes from two main sources: demographic changes (i.e. increases in the number of workers) and productivity growth (i.e. each worker producing more output). Historically, approximately 55% of growth has come from productivity growth and 45% has come from demographic changes.
  • Slowing population growth and an aging population make it unlikely that demographics will continue to be a strong tailwind to economic growth over the next 50 years.
  • Barring a productivity miracle, future economic growth is likely to disappoint those who anchor expectations to the past. Investors must adjust their expectations and strategies to fit this new reality.

Trend GDP growth is a critical determinant of traditional asset class returns.  For equities, GDP growth has historically served as an upper bound to earnings growth.  For bonds, short-term interest rates can be modeled as the sum of GDP growth, time preferences (i.e. the relative preference between saving and investment), and monetary policy effects.

When thinking about GDP growth, we prefer to follow an unbundle/re-build framework.  With this approach, we unbundle the sources of GDP growth, analyzing each both individually and in concert with the other sources of growth, and only then re-aggregate the components to top level GDP growth.

There are a number of valid methodologies for decomposing GDP growth.  In this piece, we will use a supply-side approach.  The basics are as follows.

In the equation below, let Y denote real GDP and N denote total population.  We can decompose real GDP per capita (Y / N, the amount of output/income produced on average by each member of the population) as follows:

screen-shot-2016-12-12-at-8-59-08-am

where E is total employment and L is the working age population.

The first term on the right-hand side of the equation (Y/E) reflects the amount of output generated per worker.  This is a measure of productivity.  (Note: Productivity is typically measured as output per hour worked since output per worker can change simply due to fluctuations in hours worked.  For example, increases in part-time jobs as a proportion of total jobs would reduce hours worked and output per worker even though productivity was unchanged.  That being said, long-term output per worker trends have been driven almost entirely by changes in productivity.  As a result, we are comfortable using output per worker as a proxy for productivity.)

The second term (E/L) is the proportion of the working age population that is employed.  This will be a function of the labor force participation rate (the proportion of the population that wants to work) and the employment rate (the proportion of the labor force with a job, which equals one minus the unemployment rate).

The third term (L/N) is the percentage of the population that is of working age (16 years and older).

Using a mathematical tool called the Shapley decomposition, we can represent changes in real GDP per capita as a weighted average of changes in each of these three quantities.

Once we understand the sources of per capita GDP (Y / N) growth, we can easily pivot to total GDP (Y) growth since the growth rate in per capita GDP will be approximately equal to the difference between total GDP growth and population (N) growth.  Equivalently, total GDP growth will be approximately equal to the sum of per capita GDP growth and population growth.

A Case Study: Understanding Long-Term GDP Growth in the United States (1948 to 2015)

As a case study, let’s unbundle the sources of U.S. GDP growth from 1948 to 2015.  In 1948, real GDP was $2.0 trillion in 2009 dollars.  By 2015, real GDP had growth to $16.4 trillion, an annualized growth rate of 3.2%.

Summarizing U.S. Economic Growth (1948 to 2015)

screen-shot-2016-12-12-at-9-00-16-am

Of this 3.2% growth, a little less than two-thirds (2.0%) resulted from more output per person (i.e. growth in real per capita GDP).  The remaining growth (1.2%) came simply from an increase in population.

screen-shot-2016-12-12-at-9-01-37-am

We can go a step further by using the aforementioned Shapley decomposition to attribute the per capita GDP growth to changes in productivity (output per worker), changes in the employment rate (the percentage of the working age population that is employed), and changes in population composition (the percentage of the population that is of working age).

screen-shot-2016-12-12-at-9-02-13-am

The vast majority (~89%) of per capita GDP growth can be attributed to increases in productivity.  In 1948, the average worker produced $34,613 worth of output.  This figure grew to $110,167 by 2015.  Changes in the employment rate and population composition combined to contribute just 0.23% to real per capita GDP growth.

To gain more intuition around the drivers of productivity growth, we can break it into two components:

  1. Intra-sector productivity growth (e.g. farmers producing more output per worker).
  2. Inter-sector employment shifts (e.g. workers moving from the farming sector to the manufacturing sector). If workers move from less productive sectors to more productive sectors, economy-wide productivity will increase.  If workers move from more productive sectors to less productive sectors, economy-wide productivity will decline.

screen-shot-2016-12-12-at-9-04-59-am

Productivity growth over the 1948 to 2015 period can be entirely attributed to intra-sector productivity growth.  More than half of the productivity gains came from just three sectors: manufacturing, finance, and professional and business services (which includes many tech businesses).

picture1

Data Source: Federal Reserve Bank of St. Louis. Calculations by Newfound Research. 

Inter-sector employment shifts were actually a modest detractor from productivity growth as workers on average moved from more productive to less productive industries.

The biggest negative contributors were reductions in manufacturing jobs (a relatively high productivity sector) and increases in education, health care, and social assistance jobs (a relatively low productivity sector).

The biggest positive contributors were reductions in agricultural jobs (a relatively low productivity sector) and increases in finance jobs (a relatively high productivity sector).

picture2

Data Source: Federal Reserve Bank of St. Louis. Calculations by Newfound Research. 

Putting all of this analysis together, we can list the top 10 contributors to GDP growth over the last 67 years:

  1. Population Growth (+1.18%)
  2. Productivity Growth in Manufacturing Sector (+0.51%)
  3. Productivity Growth in Finance Sector (+0.30%)
  4. Productivity Growth in the Professional and Business Services Sector (+0.17%)
  5. Productivity Growth in the Government Sector (+0.17%)
  6. Change in Population Composition (+0.16%)
  7. Productivity Growth in the Information Sectors (+0.14%)
  8. Productivity Growth in the Wholesale Trade Sector (+0.11%)
  9. Productivity Growth in the Education, Health Care, and Social Assistance Sectors (+0.10%)
  10. Employment Shift into the Finance Sector (+0.09%).

Shortening the Time Horizon

It can also be informative to narrow our time horizon and consider shorter rolling periods.  Below, we plot rolling 10-year GDP growth as well as the contribution of each of the main components discussed above.

picture3

Data Source: Federal Reserve Bank of St. Louis. Calculations by Newfound Research. 

From the graph, it’s clear that real GDP growth has been slowing, especially over the last decade.  From 1958 to 2007, the U.S. economy grew at 3%+ in 95% of rolling ten-year periods.

The economy has not returned to this growth rate since.

This decline been caused by both slowing productivity growth and unfavorable demographic developments.  From a demographic perspective, two trends have combined to form a strong headwind for growth:

Slowing overall population growth due to declining birth rates after the Baby Boomers.

picture4

Data Source: Federal Reserve Bank of St. Louis.  Calculations by Newfound Research. 

Aging of the population, leading to a decline in the employment rate – defined here as the percentage of the working age population that is employed.  From 1940 to 2010, the percentage of the population over 65 years of age nearly doubled from 6.8% to 13.1%.  Holding all else equal, the employment rate will decline as retirees make up more and more of the population since working age is defined as 16+ years of age.

picture5

Data Source: Department of Commerce (Bureau of the Census).  Calculations by Newfound Research.

Looking Forward

We can use the unbundle/re-bundle approach not only to understand the past, but also to form views on future GDP growth.

The Census Bureau publishes population and demographic projections out to 2060.  They currently project that the overall population will increase from 321 million in 2015 to 417 million in 2060, an annualized growth rate of 0.58%.

picture-correct

Data Source: U.S. Census Bureau.

We can use these same projections to estimate that 82% of the population will be of working age (16+) in 2016.  This is an increase from 78% in 2015.  If realized, this workforce expansion would contribute +0.13% to annual real GDP growth.

Unfortunately, this benefit is wiped away by the projected aging of the population.  Assuming that unemployment settles at 4.8% (the current long-term natural rate as calculated by the Federal Reserve), we project that the employment rate (workers divided by number of people of working age) will decline from 59% in 2015 to 54% in 2060.  In calculating this projection, we make the simplifying assumption that age group labor force participation rates remain constant.  This shift would be a 0.21% drag on annual real GDP growth.

screen-shot-2016-12-11-at-5-27-24-pm

The final component that needs to be forecast is productivity growth.  In our view, this is the most difficult component to forecast.  Fortunately, a precise estimate is not needed to draw meaningful directional conclusions.

Combining the impacts of the simple forecasts of population growth (+0.58%), employment level (-0.21%), and population composition (+0.13%), we estimate just 0.49% of real GDP growth from expansions in the size of the workforce.

With these estimates, any GDP growth above this 0.49% will have come from productivity growth.

This implies that for the U.S. economy to match the growth it saw from 1948 to 2015 (3.17%), productivity would have to grow by 2.68% per year.

To put this into perspective, productivity grew this rapidly in less than 7% of the rolling ten-year periods we studied.  The last time the economy achieved this feat over ten years was for the period that ended in 1968.

More realistically, we could assume that productivity growth matches the long-term average from the period we studied (despite, arguably, being a highly anomalous period due to major technological and economic innovations like the computer, the jet engine, and globalization).

From 1948 to 2015, productivity growth contributed 1.74% to annualized real GDP growth.  If the U.S. matched this accomplishment from 2015 to 2060, projected real GDP growth would be 2.23% per year.

That is a 30% discount to the long-term average.

picture7

Data Source: U.S. Census Bureau and Federal Reserve Bank of St. Louis.  Calculations by Newfound Research.

Alternatively, we could assume that productivity grows at the pace we’ve seen more recently.  Over the last ten years, productivity growth contributed 0.93% to annualized real GDP growth.  If the U.S. achieved this lower hurdle from 2015 to 2060, projected real GDP growth would be drop to just 1.42% per year, less than half of the economic growth we are accustomed to.

picture8

Data Source: U.S. Census Bureau and Federal Reserve Bank of St. Louis.  Calculations by Newfound Research.

Which of these scenarios is more realistic?  The truth is that nobody knows.  If we had to guess, we’d lean towards the more pessimistic scenario.

One main reason for this is that productivity and demographics are not independent variables.  Rob Arnott and Denis Chaves published an interesting piece on this topic (“A New ‘New Normal’ in Demography and Economic Growth”).  Quoting from the paper:

Our results show that children have a slightly negative effect on economic growth, but young adults start to positively contribute as they join the work force. Skeptics might argue that wages and productivity peak later in life, typically in one’s 40s and 50s. This is generally true, and helps to explain why the most prosperous nations often have a larger proportion of mature adults than the less prosperous nations. However, the definition of a peak, whether for productivity or anything else, is that we stop rising and start falling. When we reach peak productivity, our growth in our productivity is zero! It’s the young adults, in their 20s and 30s, who have the most rapid rate of change in their productivity. One might say that mature adults are terrific for GDP, but not for GDP growth, and that young adults are terrific for GDP growth, but less so for GDP.

The average contribution to GDP growth becomes negative between 55 and 60. Again, this does not mean that people begin to consume more GDP than they produce after age 55, only that—on average—mature workers above age 55 have passed their peak in productivity. One can readily infer from this graph that the average 60-yearold is more productive than the average 40-year-old, but not so relative to the average 55-year-old. At ages 60 and above, the coefficients decline much more sharply: The mature worker exhibits falling productivity, but in retiring, a worker’s productivity simply falls off a cliff!”

Implications for Traditional Asset Allocation

We expect low economic growth to be a drag on traditional stock and bond returns and we are not alone.  For example, Research Affiliates expects a domestic 60/40 stock/bond portfolio to return just 1.2% per year over the next decade after inflation.  This pessimism results from the triple whammy of low growth, low current yields, and high valuations.

In response, we advocate for a four-pronged response:

  1. Set expectations that are consistent with the current reality, not the multi-decade bull market we’ve seen in both stocks and bonds over past decades.
  2. Seek return opportunities beyond U.S. stocks and bonds.  Specifically, this means looking for higher yields, better growth prospects, and more favorable valuations.  We see this play out in Research Affiliates’ capital market assumptions.  Below, we show their return-optimal portfolio with expected volatility that matches the traditional 60/40 stock/bond mix.  We see a large foreign equity allocation (where valuations are more reasonable, yields are higher, and growth prospects are moderately better), an underweight to traditional fixed income, and significant allocations to alternative asset classes, especially those that offer high income like high yield and bank loans.
  3. Institute a robust risk management plan that acknowledges the lessons we’ve learned in behavioral finance and looks beyond bonds for risk mitigation (e.g. complementing bonds with allocations to actively risk-managed strategies like tactical asset allocation, managed futures, and equity long/short).
  4. Look for areas where returns may be incrementally increased through low cost, evidence-based investment (e.g. factor-based equity strategies).

picture9

Source: Research Affiliates.  Allocations are based on the forward-looking beliefs of Research Affiliates, are not to be considered investment advice, and are not a guarantee of future performance.

Conclusion

Trend GDP growth is a critical determinant of traditional asset class returns.  For equities, GDP growth has historically served as an upper bound to earnings growth.  For bonds, short-term interest rates can be modeled as the sum of GDP growth, time preferences (i.e. the relative preference between saving and investment), and monetary policy effects.

When thinking about GDP growth, we prefer to follow an unbundle/re-bundle framework.  With this approach, we unbundle the sources of GDP growth, analyzing each both individually and in concert with the other sources of growth, and only then re-aggregate the components to top level GDP growth.

We can use this approach to both understand historical and project future economic growth.  Historically, approximately 55% of growth has come from productivity growth and 45% has come from demographic changes.  More than half of productivity growth came from three sectors: manufacturing, finance, and business and professional services (which includes many tech businesses).

Going forward, we expect demographics to contribute about 0.5% to annual GDP growth.  As a result, future growth is likely to lag past growth even if productivity growth accelerates out of its current lull and back to the historical average.

Client Talking Points

  • Economic growth is a key driver of long-term stock and bond returns.
  • Economic growth comes from two main sources: demographic changes (i.e. increases in the number of workers) and productivity growth (i.e. each worker producing more output). Historically, approximately 55% of growth has come from productivity growth and 45% has come from demographic changes.
  • Slowing population growth and an aging population make it unlikely that demographics will continue to be a strong tailwind to economic growth over the next 50 years.
  • Barring a productivity miracle, future economic growth is likely to disappoint those who anchor expectations to the past.  Investors must adjust their expectations and strategies to fit this new reality.

 

Justin is a Managing Director and Portfolio Manager at Newfound Research, a quantitative asset manager offering a suite of separately managed accounts and mutual funds. At Newfound, Justin is responsible for portfolio management, investment research, strategy development, and communication of the firm's views to clients.

Justin is a frequent speaker on industry panels and is a contributor to ETF Trends.

Prior to Newfound, Justin worked for J.P. Morgan and Deutsche Bank. At J.P. Morgan, he structured and syndicated ABS transactions while also managing risk on a proprietary ABS portfolio. At Deutsche Bank, Justin spent time on the event‐driven, high‐yield debt, and mortgage derivative trading desks.

Justin holds a Master of Science in Computational Finance and a Master of Business Administration from Carnegie Mellon University as a well as a BBA in Mathematics and Finance from the University of Notre Dame.