Wednesday, June 09, 2021

On Business Cycles


I'm in the middle of another class this week and for the past week, I've been trying to crack one of the best books ever written on market cycles. Lars Tvede clearly hit the ball right out of the park with his bible Business Cycles and I strongly urge serious investors to read this book as it is clear and actually gives the reader a chance to quantitatively chart the business cycle if they can have access to economic data.

This book is extremely broad and covers the history of myriad attempts at defining the cycle as well as what is experienced for investors on the ground. My only regret is that it is impossible to summarise the book for my students. 

I probably need a few months to engineer a tool to create a business cycle for Singapore. 

A few useful lessons I picked up from the book :

a) It's not one cycle, but at least three cycles working in tandem

The best way to understand business cycles is that it is a combination of at least three different processes. The Kitchen's cycle is based on inventory and this has a period of about 4-5 years. The second cycle which is the Juglar cycle is based on capital expenditure that rises and falls and this can have a period of about 9 years. The longest cycle or the Kurznet's Cycle and can run as long as 18 years. 

To complicate matters even further there is a rumoured Kondratieff cycle that runs for half a century.

If you take this idea seriously, it would be very challenging to break the actual cycle into different phases. But it can be rewarding because avoiding a property crash is probably the aim of all serious investors.

b) There is no consensus over what measurement to use within a cycle

The next problem is to figure out which metrics can be used to trace a cycle. High rental rates may discourage inventory and not all inventory plays a large role. Somehow vehicle inventories play a larger role in determining this cycle. 

The same applies to capital spending cycles, there is a theory that shipbuilding rates play a big role and some ratios involving CAPEX in the local economy needs to be factored in as well. 

Property prices also involve complex metrics like building a ratio of CAP rate to bond yields. 

If I want to engineer my own business cycle, will need to gather data and build my own index. 

c) After you have engineered what seems to be a business cycle time series, it must be able to predict asset price trends based on historical data

Once you can start modelling a business cycle, it must be able to predict asset prices. It should ideally be able to lead asset price trends by a few months. 

If a business cycle trend is built, ideally it should be able to predict the downwards trend in stocks when it begins to decline from its peak. Whether this works or not can be covered by a regression exercise or any machine learning tool. 

These are all fairly tough problems, but I think the book has given confidence that coming up with a business cycle using local data is not impossible. I'm pretty sure some econometrics major would have already succeeded in this endeavour. You may wish to take inspiration from the RICI index developed by Jim Rogers.

It may be a while before I can digest enough to make an impact on future training materials. 

I will update this blog on Saturday evening when my program concludes.


  1. Econ/biz cycle investing books were popular in the 1990s & 2000s. It evolved into sector rotation.

    For US, easy method is to simply check the 10 or 12 sector ETFs with a 6 month lookback on a monthly basis, and put your money into the top 2 or 3. These shouldn't change that often.

    For long or secular cycles, the main problem is being able to stick with the intra-downturns along the way. Most of the time, unless it's a buy-and-hold retirement account, we'll be doing lots of messing around when prices drop by 20% or 40%.

    Funny that you brought up the RICI indices --- they were developed right at the top of the last commodity secular cycle. A bit like the magazine cover indicator lol.