The full title of this book is “*Unexpected Returns: Understanding Secular Stock Market Cycles*” by Ed Easterling.

How I came across this book in the first place was when I was looking into the relationship between stock market returns and a number of other factors, such as interest rates, inflation, earnings, dividend yield, unemployment, GDP, home price index, etc. I pulled down the raw data from here, plugged them into Excel and looked at the various relationships. After some time, I saw that there is a consistent relationship between stock price levels and the inflation rate. The stock market does well in times of low, stable inflation, and does poorly in times of high inflation or deflation. I then googled around to see if others have done any research on the relationship between the stock market and inflation, and had the good fortune to find a wonderful chart on Ed Easterling’s website: Crestmont Research with the same results!

The great thing was, Ed’s website contained many other interesting results from his research. There and then, I knew I found a treasure trove, plus a person that does interesting research. I actually did not have much time to read through his articles on the website, but I learned that he had published two books, the first of which is *Unexpected Returns*. I finally got a copy of this and quickly digested it over 3 days.

First off, this is a great book. The Financial Physics model linked together the concepts of P/E ratio, inflation, GDP, and earnings growth in a coherent, logical framework. The model allows one to forecast the long term trend of the stock market and its returns. This is one model all investors should definitely have in their toolbox. The research done was good, and I managed to replicate a number of his results from my own data. The explanations were also clear and logical, though one gripe is that the explanations on the Financial Physics model were repeated several times in different parts of the book. His Y-Curve graph probably appeared 3 times in the book.

One interesting output of the Financial Physics model is a scenario analysis of future stock market returns based on inflation and P/e ratio assumptions. I liked the fact that the Financial Physics model allowed a number of unlikely scenarios to be removed which allowed the likely scenarios to be narrowed down and evaluated.

One great thing about this book is that it is printed in colour, on good quality paper for all pages. That really helps for the various graphs and tables, and makes reading the book much more enjoyable. While that must have increased the cost of the book, I am glad the price of the book remained reasonable. Another good thing is that most of the charts in the book are also found on Ed’s website.

In terms of application, my thoughts on the Financial Physics model is that while it is good for long-term projections, it is difficult to use for market timing. The case in point is that the book uses data around 2003 and concluded that it may be the early stages of the secular bear market, however, P/E ratios remained high over a number of years, and the downturn only came in 2008. What came to my mind was, how does this model gel with the methodology in Buffettology, where Buffett purportedly simply uses an average P/E multiple as the terminal exit multiple after 5-10 years? That would essentially throw away the need for any projections, and simply make the assumption that the P/E multiple will revert back to the mean within 5-10 years. Essentially, using the Financial Physics model, it implies that within 5-10 years, inflation should fluctuate in such a way that causes the P/E to revert to its mean. To me, that sounds like a perfectly reasonable assumption, and does away with the need for market timing or stock market projections.

Key points in the book follows:

**Secular Cycles Correspond to Increasing / Declining P/E ratios**

- Secular bull periods are typified by generally rising P/E ratios, and secular bear periods are typified by generally declining P/E ratios (here).
- Secular market periods range from 4-24 years.
*P/E ratio at the start of the investment, is negatively correlated with annual total return over 20 years*(here and here). The change in P/E ratios is a main driver of returns. Looking at returns over twenty-year rolling periods (85 periods), periods with the best returns generally start with low P/E ratios and end with high P/E ratios, and conversely for periods with the lowest returns (here).- The distribution of returns in secular bull and bear periods are very different.
- Even in secular bear periods, there are years with annual returns of > 10% though the proportion is much less. However in secular bull periods, the frequency of years with < negative 10% returns, is extremely small (here).
- Nonetheless, the number of up days to down days is close to 50/50 (up days slightly more) across secular cycles, decades, and individual years (here).
*[My note: this actually means that bull and bear secular cycles are generated due to the magnitude of returns in the up and down days, rather than the number of up and down days.]* - Breakdown of returns in the secular bull period of 1982 – 1999
- For S&P500 Index, annual return of 17%, comprising of 8% due to P/E increase, 3.1% due to Dividend Yield, 5.9% due to EPS Growth. With average inflation of 3%, real return was 14% p.a.
- For 20-Year Treasury Bond, annual return of 12%, comprising 3.6% due to decline in interest rates, and 8.3% due to the interest payment yield. Real return of 9%.

**Inflationary / Deflationary Trends Drive Changes in P/E ratios**

- Periods of rising P/E ratios correspond to periods in which inflation moves toward price stability (low inflation).
- Periods of falling P/E ratios correspond to periods in which inflation moves away from price stability, either toward greater inflation or greater deflation.
- Easterling’s Y-Curve illustrates the above in a scatter plot showing that high inflation or deflation leads to low P/Es while stable inflation (~3%) leads to high P/Es (here).
- High inflation drives down P/E ratios because rising inflation leads to rising interest rates, which increases investors’ required return, which drives down prices (here).
- Deflation drives down P/E ratios because earnings and dividends (both current and expected future) decline significantly, which drives down prices. Even though interest rates would be low, it is not sufficient to push up the P/E ratios.
*[My note: while cyclical stocks tend to have high P/E ratios when their earnings reach the bottom, expectations on their future earnings can still be reasonably decent which keeps the price (and correspondingly the P/E) high. However in a deflationary scenario, future earnings projections would be much more severely impacted (negative EPS growth), leading to lower prices and P/Es).*

**Stock Market Levels are Not Explained by **Economic Growth

- Historically, nominal GDP grew faster on average during bear markets than during bull markets (here).
- Periods of high nominal GDP growth typically results from high inflation, which lowers the P/E ratio.
- While economic growth increases earnings of companies, the P/E ratio is still the main driver of actual returns.

**Stock Market Levels are Not Explained by Interest Rates**

- During periods of high inflation, interest rates rises and lowers market P/E ratios (here).
- However during periods of deflation, interest rates fall, yet P/E ratios also fall. This is because future earnings are expected to decline so prices fall as well.

**Inflation and Interest Rates are on the Same Roller-Coaster only after 1960s**

- Relationship (here)
- Interest rates rise when inflation increases because additional interest is required to compensate for the loss of buying power due to inflation.
- Why relationship wasn’t consistent prior to 1960
- Previous inflationary/deflationary periods were seen as temporary (e.g. after wars) hence interest rates were not used as a tool to address these issues.
- Interest rates are more volatile than most people think
- Easterling’s 6/50 Rule (here) states that during every 6-month period, interest rates have moved by more than 50 bps somewhere along the yield curve. This has happened for every single rolling 6-month period from 1965 to 2003 except for 2 exceptions (over 99% of the time).
- Higher prices are not necessary inflation
- They can be due to better quality, quantity (e.g. bigger homes cost more), higher demand, lower supply, etc. Inflation is a broad-based increase in prices.
- Two ways to measure inflation
- GDP deflator: Reported by the Bureau of Economic Analysis (BEA). The deflator for the base year is set to 100 with other years reported relative to the base year’s dollar. E.g. if the base year is 2005, and the deflator for 2008 is 108, it means that inflation in the GDP was 8%, so you need 8% more dollars in 2008 than in 2005 to buy the same amount of goods.
- % Change in CPI: ’nuff said.

**Predicting the Future using the Financial Physics Model (here)**

The Model

- Use inflation expectation to forecast the future P/E ratio
- Forecast Nominal GDP growth and use that to forecast the future EPS
- Multiply the future P/E ratio and future EPS to get the future stock market level

Forecasting Stock Market Level

- Step 1: Forecast Real GDP Growth
- From 1900 – 2003, Real GDP growth averaged 3.5% per year, with most decades between 3% and 4.5%.
- During the last 3 decades (1970s, 1980s, 1990s), Real GDP growth averaged ~3% to 3.2% per year.
- Since the historical Real GDP growth is consistent, we can use 3% to project forward.
*[My note: I actually found it odd that Easterling used annual simple averages in his calculations of historical GDP growth and inflation, despite warning earlier that compounded returns should be used instead of simple average returns)].*- Step 2: Forecast Inflation and Nominal GDP Growth
- From 1900-2003, average Real GDP growth = 3.5%, Nominal GDP growth = 6.7%, inflation in the GDP = 3.1%, EPS growth = 6.1%.
- From 1900-2009, average Real GDP growth = 3.3%, Nominal GDP growth = 6.3%, inflation in the GDP = 2.9%, EPS growth = 4.3%.
- From 1950-2009, average Real GDP growth = 3.3%, Nominal GDP growth = 6.8%, inflation in the GDP = 3.5%, EPS growth = 5.3%.
- From 1900-1949, average Real GDP growth= 3.3%, Nominal GDP growth = 5.6%, inflation in the GDP = 2.2%, EPS growth = 3.2%.
- Use average inflation of 3.2% and Real GDP growth of 3% to predict average future Nominal GDP growth of 6.2%.
- Step 3: Forecast EPS
- Method 1: Regress with Nominal GDP
- Nominal GDP and EPS are highly correlated (correlation coefficient ~0.95).
- Use the Nominal GDP growth rate above to estimate future Nominal GDP.
- Regress historical Nominal GDP versus EPS and use the equation to estimate future EPS.
- The estimated future EPS is a centerline around which EPS can be expected to cycle over time.
- This works because the Nominal GDP is effectively the combined revenues of all companies. Earnings is derived from revenues, and tend to grow at roughly the same rate as revenues. The growth rate of earnings for S&P500 companies would be slightly slower than the overall economy because the economy consists of other fast-growing small businesses.
*[My note: I found that regressing ln(EPS) vs ln(Nominal GDP) gives an amazing r-squared of ~97%. The equation obtained is ln(EPS) = 0.8197 * ln(Nominal GDP) – 9.5611. With this equation, 2010 EPS has overshot the normalized EPS by quite a bit (77.35 vs. 52.30), which forecasts that EPS will drop significantly in the near future. Timing is always a tough call because the chart also showed that actual EPS was above the normalized EPS from 2003 – 2007. I also found that if you regress EPS vs. Nominal GDP, you get an r-squared of ~90% which is worse, but the predicted normalized EPS is higher and closer to the actual EPS. I would go with the model that has the higher r-squared.]*- Method 2: Historical growth rate
- EPS growth rates ‘cycle’ around with positive growth rates for 3-5 years, followed by negative growth rates for 1-3 years. From 1950-2010, the 3-year average EPS growth rate is 5.6% (here).
- Method 3: Profit margin
- Another way to model future EPS is to note that Pre-Tax Corporate Profits as % of GDP has historically averaged ~9.1% from 1929-2010 .
*[My note: you can get raw data from here. The EPS growth rate can be estimated as (1 + GDP growth rate) * (1 + expected % change in pre-tax profit margin)]*- Step 4: Forecast P/E
- Use the Y-Curve to obtain the corresponding P/E for the assumed inflation rate (here).
- For an inflation assumption of 3.2%, the P/E ratio is 16.9.
- Step 5: Forecast Stock Market Index Level
- Multiply the forecast EPS with forecast P/E ratio.

Forecasting Stock Market Returns

- Total Return = Market Price Change + Dividend Yield
- Market Price Change = % change in Earnings + % change in P/E ratio
- Dividend Yield can be forecast given a P/E ratio by using this curve here. Alternatively, dividends have averaged 50% of earnings.
- This here shows the potential stock market return based on the inflation and P/E ratio assumptions.
- Note that the EPS growth is driven by inflation because inflation drives Nominal GDP growth, which is slightly higher than EPS growth.
- Dividend yield is also driven by the P/E ratio as explained above.
- There are a number of high P/E ratios corresponding to 1-2% inflation because from the Y-curve, 1-2% inflation does indeed correspond to high P/E ratios.
- The book “blanked out” some unlikely scenarios which the website document didn’t. Scenarios that were removed are where the EPS Growth was either too high or too low, relative to the inflation assumption, i.e. (i) where EPS Growth > (Real GDP Growth of 3% + Inflation assumption), and (ii) where EPS Growth < (Real GDP Growth of 3% + Inflation assumption – ~2%).
- Some inconsistent scenarios are not shown, e.g. P/E ratios are high and inflation is high, which is not consistent because P/E ratios are low when inflation is high.

**Volatility Drives Out Investors, Hurting Returns**

- The more volatile the market, the lower the expected return (here). Market volatility is measured as a average of [day’s range (high – low) / day’s starting value].
- Volatility drives investors out of their stock holdings, so investors would realize the decline but not be invested during the recovery.
- If you consider different scenarios, all with the same average return, the greater the dispersion of the returns (i.e. the wider the range) from the average, the lower the compounded return (here). Investors can only spend compounded returns, not average returns.

**P/E ratios have Natural Limits**

- The dividend discount model can be used to derive natural limits to the P/E ratio (here).
- Inputs
- Required rate of return = Treasury bill interest rate + spread to bring it to Treasury Bond yield + spread to bring it to Corporate Bond yield + spread to bring it to Stock Market Return. The inflation assumption would impact the starting Treasury bill interest rate, the corporate bond spread, and the stock market return spread.
- EPS growth rate = ~0.4% to 0.8% less than the Nominal GDP growth rate (fixed Real GDP growth rate of 3% + inflation assumption)
- Dividend payout ratio = 50%.
- Results for natural P/E ratios
- -3% inflation scenario: P/E = 9x
- 1% inflation scenario: P/E = 25x
- 2% inflation scenario: P/E 20x
- 3% inflation scenario: P/E = 17x
- 5% inflation scenario: P/E = 13x

**How to Assess the Current Market Cycle**

Look at the following factors:

- P/E ratio
- Historical average P/E since 1900 is ~15.5x to 16x (here). The normal range is around ~10x to 30x (note these are Shiller’s P/E 10 values).
- If the P/E ratio is well above average, further increases would be unlikely.
*[My note: It is interesting that at the time of the book (around end of 2004?), Easterling estimates that the P/E ratio is already high with respect historical data, and would unlikely go higher. Yet as it played out, P/E went from 23x in 2003 to close to 28x in 2004, maintained above 25x until 2008. In the book Easterling predicts that it may be the start of a secular bear market, but it turned out quite differently. It just shows how hard it is to get the timing right.]*- Dividend yield
- Historical average dividend yield since 1900 is 4.4% (others report the average at ~3.5% to 4%).
- The amount of dividends are determined by profits and internal capital needs, which are independent of the P/E ratio. Hence low/(high) valuations in the market would imply a high/(low) dividend yield (see here) since the payout ratio of dividends to earnings historically has been consistently 35% to 60%, with an average of ~50%.
- The dividend yield can be used as a measure of the valuation level of the market instead of the P/E ratio because it is generally more stable. Earnings are impacted by temporary adjustments or one-time charges, which impacts the P/E but usually not the dividend yield.
- Inflation
- Historical average since 1900 is 3.3%. Look at the potential inflation trend.
- Interest Rates
- Historical average since 1900 of the 20-Year Treasury Bond Yield is 4.9%, the 40-year average is 8.1%.
- Low interest rates would reflect conditions of relatively high valuation.

**Investment Implications**

- When in the early stages of a secular bear market, switch to absolute return strategies.
- When in the early stages of a secular bull market, switch to relative return strategies that track the Index benchmark.

**Investment Strategies**

For Stocks:

- To increase returns, rebalance more frequently in a secular bear market, and rebalance less frequently in a secular bull market (here).
- If you want a consistent strategy without caring about the type of market, rebalance more frequently. The benefits during bear markets outweigh the cost during bull markets.

For Bonds

- Bond investors benefit from two areas:
- Interest payment from the bond
- Increase in the bond value as the bonds “roll down the yield curve”, i.e. as the maturity decreases, the bond value increases because shorter-term yields are lower in a positive sloping yield curve.

- Use a Bond Ladder
- Portfolio of bonds, with a portion maturing each year. The cash from the matured bond is used to buy new bonds. E.g. a 10-year bond ladder might have 10 bonds with maturities 1 to 10 years, and 10% invested into each bond. Every year, the cash from the bond that matures will be used to buy another 10-year bond.
- Why It Works
- Receive higher interest from longer-term bonds
- Discipline of the strategy avoids market-timing
- In periods of rising interest rates, you consistently get to buy cheaper bonds.
- In periods of declining interest rates, you are receiving the higher yields from the long-term bonds bought earlier.
- In either case, you continue to receive the benefits of interest payments and increased value due to the roll.
- How to handle rising interest rates
- To assess whether to postpone buying bonds and instead sit in money money funds, determine the “break-even yield”. Take for example a 10-year
**par**bond with a 1-year holding period: - If you don’t buy the bond, you get (A) principal + principal * money market interest rate
- If you buy the bond, and interest rates rise, you get (B) value of the bond with the 9-year yield (which rose) + principal * current 10-year yield
- The break-even yield is the 9-year yield that equates (A) and (B), i.e. the future yield at which the current yield needs to rise to, in order to make the bond not worth buying. You can similarly calculate the break-even yields for other maturities, or for different holding periods.
- Finally, assess how likely is it for the current yield to rise to the break-even yield for the particular holding period.

You have a link for the FRED which contains in its website the Condition of Banks for U.S. banks. Do you have another link to download data for the Condition of Banks for European banks? Thanks.

Hello,

I have not seen a single source on EU banks that is similar to the FRED source for U.S. banks. If you have access to Bloomberg, Capital IQ, or Reuters, you should be able to get the same information for the major EU banks. If you are looking for public sources, there are some statistics here:

1) High level statistics: http://www.ebf-fbe.eu/index.php?page=statistics

2) BIS data: http://www.bis.org/statistics/bankstats.htm

3) ECB macro data: http://www.ecb.int/stats/html/index.en.html

Do share if you do find some better sources. Hope this helps.

Hi there, thanks for sharing your excellent review deconstructing/reconstructing Ed Easterling’s model. Do you actually use this approach to forecast S&P500 price targets? Do you think it can be applied successfully?

Hi Chris, I don’t use this approach nor do I forecast S&P500 price targets. There are a number of difficulties with applying this approach: (i) need to forecast the inflation rate and real growth rate, (ii) even if you get those right, you may have to sit through some decent short-term volatility which can mess with your mind, (iii) you are still left with the problem of selecting what stocks to play. I think the value of Easterling’s framework is more to identify where we are sitting within the secular market cycles. It gives a more long-term perspective rather than something to be used for short to intermediate term tactical investing (which fits more with my style).