Testing unemployment models for other countries besides the U.S.

Troy Bombardia
2019-10-29
null

Applying the U.S. unemployment rate to a simple trend following strategy allows us to better time the S&P 500.

  1. Buy and hold the S&P 500 (nontotal return index) if the S&P 500 is >= its 12 month moving average, OR the Unemployment Rate is <= its 12 month moving average
  2. Otherwise, SELL

Can we use this same concept in other countries' stock markets? I.e. is this strategy timeless and universal, or does it only apply to the U.S.? I ran the following strategy on 4 different countries' stock indices:

  1. DAX (Germany)
  2. S&P/ASX 200 (Australia)
  3. S&P/TSX Composite Index (Canada)
  4. Nikkei 225 (Japan)

The strategy is as follows:

  1. Buy and hold the stock index if the index is >= its 12 month moving average, OR the country's Unemployment Rate is <= its 12 month moving average.
  2. Otherwise, SELL.

You can do 1 of 7 things when you sell:

  1. Shift into 100% cash
  2. Short the stock index
  3. Buy and hold the Bloomberg Barclays US Aggregate Bond Total Return Index
  4. Buy and hold the Bloomberg Barclays US Corporate Bond Total Return Index
  5. Buy and hold the Bloomberg Barclays US Treasury Bond Total Return Index
  6. Buy and hold the Bloomberg Barclays 1-5 Year US Treasury Bond Total Return Index
  7. Buy and hold gold

*I would prefer to use the country's own bond index instead of a U.S. bond index, but historical data for ex-U.S. bond indices is limited. Investing in ex-U.S. stock indices + U.S. bond indices also introduces the problem of currency conversions, which I have not included in this study. This study is purely meant to demonstrate the usefulness (or lack of) of unemployment as a long term timing indicator.


DAX

Here's Germany's unemployment rate:

  1. January 1970 - September 2019: Model (cash) = an average of 4.29% per year
  2. January 1970 - September 2019: Model (long/short) = an average of 2.27% per year
  3. January 1976 - September 2019: Model (U.S. Aggregate Bonds) = an average of 7% per year
  4. January 1973 - September 2019: Model (U.S. Corporate Bonds) = an average of 7.08% per year
  5. January 1973 - September 2019: Model (U.S. Treasury Bonds) = an average of 6.58% per year
  6. January 1976 - September 2019: Model (U.S. 1-5 Year Treasury Bonds) = an average of 6.61% per year
  7. January 1970 - September 2019: Model (Gold) = an average of 5.78% per year

*Historical returns do not take into consideration dividends reinvested, different currencies & currency fluctuations, transaction costs, slippage, etc.

Since the above chart is hard to read (too many equity curves), the following chart only demonstrates the DAX vs. the Model (U.S. Treasury Bonds):

Max Drawdowns (using monthly CLOSE $):

  1. Model (cash) = -52.78%
  2. Model (long/short) = -66.79%
  3. Model (U.S. Aggregate Bonds) = -52.35%
  4. Model (U.S. Corporate Bonds) = -52.55%
  5. Model (U.S. Treasury Bonds) = -52.78%
  6. Model (U.S. 1-5 Year Treasury Bonds) = -52.4%
  7. Model (gold) = -55.09%

Since the above chart is hard to read (too many drawdown curves), the following chart only demonstrates the Model (U.S. Treasury Bonds)'s drawdowns: 

Max Drawdown / Average annual return ratio:

  1. Model (cash) = 12.28
  2. Model (long/short) = 29.3
  3. Model (U.S. Aggregate Bonds) = 7.46
  4. Model (U.S. Corporate Bonds) = 7.41
  5. Model (U.S. Treasury Bonds) = 8.01
  6. Model (U.S. 1-5 Year Treasury Bonds) = 7.92
  7. Model (gold) = 9.52


S&P/ASX 200

Here's Australia's unemployment rate:

  1. May 1993 - September 2019: Model (cash) = an average of 5.49% per year
  2. May 1993 - September 2019: Model (long/short) = an average of 5.74% per year
  3. May 1993 - September 2019: Model (U.S. Aggregate Bonds) = an average of 6.02% per year
  4. May 1993 - September 2019: Model (U.S. Corporate Bonds) = an average of 6.16% per year
  5. May 1993 - September 2019: Model (U.S. Treasury Bonds) = an average of 5.9% per year
  6. May 1993 - September 2019: Model (U.S. 1-5 Year Treasury Bonds) = an average of 5.86% per year
  7. May 1993 - September 2019: Model (Gold) = an average of 6.31% per year

*Historical returns do not take into consideration dividends reinvested, different currencies & currency fluctuations, transaction costs, slippage, etc.

Since the above chart is hard to read (too many equity curves), the following chart only demonstrates the S&P/ASX 200 vs. the Model (U.S. Treasury Bonds):

Max Drawdowns (using monthly CLOSE $):

  1. Model (cash) = -29.27%
  2. Model (long/short) = -37.65%
  3. Model (U.S. Aggregate Bonds) = -28.04%
  4. Model (U.S. Corporate Bonds) = -29.67%
  5. Model (U.S. Treasury Bonds) = -28.28%
  6. Model (U.S. 1-5 Year Treasury Bonds) = -28.55%
  7. Model (gold) = -31.66%

Since the above chart is hard to read (too many drawdown curves), the following chart only demonstrates the Model (U.S. Treasury Bonds)'s drawdowns: 

Max Drawdown / Average annual return ratio:

  1. Model (cash) = 5.32
  2. Model (long/short) = 6.56
  3. Model (U.S. Aggregate Bonds) = 4.65
  4. Model (U.S. Corporate Bonds) = 4.81
  5. Model (U.S. Treasury Bonds) = 4.79
  6. Model (U.S. 1-5 Year Treasury Bonds) = 4.86
  7. Model (gold) = 5.01


S&P/TSX Composite Index 

Here's Canada's unemployment rate:

  1. January 1961 - September 2019: Model (cash) = an average of 5.18% per year
  2. January 1961 - September 2019: Model (long/short) = an average of 4.85% per year
  3. January 1976 - September 2019: Model (U.S. Aggregate Bonds) = an average of 9.97% per year
  4. January 1973 - September 2019: Model (U.S. Corporate Bonds) = an average of 9.39% per year
  5. January 1973- September 2019: Model (U.S. Treasury Bonds) = an average of 8.81% per year
  6. January 1976 - September 2019: Model (U.S. 1-5 Year Treasury Bonds) = an average of 9.41% per year
  7. January 1961 - September 2019: Model (Gold) = an average of 6.19% per year

*Historical returns do not take into consideration dividends reinvested, different currencies & currency fluctuations, transaction costs, slippage, etc.

Since the above chart is hard to read (too many equity curves), the following chart only demonstrates the S&P/TSX Composite Index vs. the Model (U.S. Treasury Bonds):

Max Drawdowns (using monthly CLOSE $):

  1. Model (cash) = -47.01%
  2. Model (long/short) = -62.86%
  3. Model (U.S. Aggregate Bonds) = -39.16%
  4. Model (U.S. Corporate Bonds) = -39.16%
  5. Model (U.S. Treasury Bonds) = -39.16%
  6. Model (U.S. 1-5 Year Treasury Bonds) = -39.27%
  7. Model (gold) = -52.76%

Since the above chart is hard to read (too many drawdown curves), the following chart only demonstrates the Model (U.S. Treasury Bonds)'s drawdowns:

Max Drawdown / Average annual return ratio:

  1. Model (cash) = 9.06
  2. Model (long/short) = 12.95
  3. Model (U.S. Aggregate Bonds) = 3.93
  4. Model (U.S. Corporate Bonds) = 4.2
  5. Model (U.S. Treasury Bonds) = 4.48
  6. Model (U.S. 1-5 Year Treasury Bonds) = 4.17
  7. Model (gold) = 8.52


Nikkei

Here's Japan's unemployment rate:

  1. January 1971 - August 2019: Model (cash) = an average of 5.49% per year
  2. January 1971 - August 2019: Model (long/short) = an average of 6.17% per year
  3. January 1976 - August 2019: Model (U.S. Aggregate Bonds) = an average of 7.38% per year
  4. January 1973 - August 2019: Model (U.S. Corporate Bonds) = an average of 6.62% per year
  5. January 1973 - August 2019: Model (U.S. Treasury Bonds) = an average of 6.34% per year
  6. January 1976 - August 2019: Model (U.S. 1-5 Year Treasury Bonds) = an average of 6.98% per year
  7. January 1971 - August 2019: Model (Gold) = an average of 5.49% per year

*Historical returns do not take into consideration dividends reinvested, different currencies & currency fluctuations, transaction costs, slippage, etc.

Since the above chart is hard to read (too many equity curves), the following chart only demonstrates the Nikkei 225 vs. the Model (U.S. Treasury Bonds):

Max Drawdowns (using monthly CLOSE $):

  1. Model (cash) = -64.62%
  2. Model (long/short) = -59.08%
  3. Model (U.S. Aggregate Bonds) = -39.63%
  4. Model (U.S. Corporate Bonds) = -38.54%
  5. Model (U.S. Treasury Bonds) = -39.86%
  6. Model (U.S. 1-5 Year Treasury Bonds) = -41.31%
  7. Model (gold) = -66.37%

Since the above chart is hard to read (too many drawdown curves), the following chart only demonstrates the Model (U.S. Treasury Bonds)'s drawdowns: 

Max Drawdown / Average annual return ratio:

  1. Model (cash) = 11.76
  2. Model (long/short) = 9.57
  3. Model (U.S. Aggregate Bonds) = 5.36
  4. Model (U.S. Corporate Bonds) = 5.81
  5. Model (U.S. Treasury Bonds) = 6.27
  6. Model (U.S. 1-5 Year Treasury Bonds) = 5.91
  7. Model (gold) = 12.08


Conclusion

The positive: Using a country's unemployment rate and 12 month moving average to time the market might not necessarily beat buy and hold, but it rarely lags buy and hold. 

The negative #1: this strategy doesn't always beat buy and hold, particularly if the country's unemployment rate consistently goes up when there is no recession and corporate profits don't decline. In such a case, the unemployment part of this model will be on a SELL signal (since unemployment is trending up). As long as the stock market falls, the 12 month moving average part of this model will be on a SELL signal as well, flipping the whole model into a SELL signal. But if the stock market becomes choppy, your portfolio will be hurt by "death by a thousand cuts".

The negative #2: In order for this strategy to turn bearish, the unemployment rate needs to trend up and the stock market needs to trend down. Since this is a 2-part model, the model will change to a SELL signal too late if the stock market peaks long before unemployment bottoms. In other words, there is no guarantee that the relationship between unemployment (economy) and stocks will hold in the future.

Sorry, you don't have access to this report

Upgrade your subscription plan to get access