Wednesday, October 13, 2021

Is CPF really the equivalent of a 60:40 fund ?

Sharp readers from BIGS forum and this blog have raised an issue on my last article which claimed that the CPF was equivalent to a 60:40 fund. The readers were rightfully incredulous. How can the CPF program which returned a fixed amount of 2.5%-4.0% be the equivalent of a 60/40 fund?

I went back to trace the origins of the reference and found this link. You can take a crack at reading this paper.

As it turns out the readers were right. The original paper was full of spurious assumptions that do not gel with our observed reality. 
  • Singapore cash rates to be as high as 4% from 2014 - 2024.
  • Singapore government bond yields to be high as 5.3% over the same period of time.
  • Global equities to return 5% above inflation over that same period of time. 

The collective effect of these assumptions was that it overestimates returns of cash rates and in turn, CPF returns as it is tied to bond yields once it is above the floor, but it also downplays equity returns - I recall Professor Jeremy Siegel of Wharton offering a different rule of thumb suggesting that equities return 6.8% over inflation. 

Once the right assumptions are in place, any conclusion can be made. 

The paper goes on to construct an efficient frontier and magically lines up CPF returns horizontally against a 60:40 portfolio. 

It then concludes that Singaporeans get the best deal because they get 60/40 returns at low risk. 

Now, I want to take it one step further. 

I was able to repurpose my Python code that produces efficient frontiers and I can extract actual historical data for the past 20 years for an equity (VT) and bond (BNDW) ETF. Using the same methodology, I pegged my CPF returns as a sole proprietor who performs an actual voluntary contribution spread over CPF-OA, SA and RA. 

My returns, not counting my tax benefits, are fixed at 3.3%.  

The resulting diagram from my program is as follows :


This makes more sense in practice, there is no way CPF returns can line up against an efficient frontier of a real global equity and bond ETF. But that's not a big deal given that CPF investments are risk-free.

All these raises issue that vindicates what a reader "Unknown" has said. The truth lies between Terence Ho and Teo You Yenn. For every left-winged fanatic hell-bent on raising taxes, there may be a staunch government apologist out there with their own "alternative facts".

There are, of course, other issues that raise eyebrows when we study the paper further. 

There are very credible academics in our local universities like NUS, why invite four westerners to author the paper under this entity called Towers Watson? 

I guess the next thing to do is SOP for this blog. 

You have friends in CPF. Tell CPF staff to read my articles and maybe even explain this strange working paper. 




5 comments:

  1. So the book author relied on 3rd party "massaged" sources rather than confirming with actual primary data or sources? Easy way to get yourself debunked.

    I suspect the same with his Gini assertion between S'pore & UK, when the official numbers show the opposite.

    Even with so-called primary sources, we need to be discerning e.g. data coming from China, Russia, and many 3rd world countries.

    ReplyDelete
  2. My guess is the probability is lower as Singapore's numbers still make it no. 2 on the list. But no harm verifying. The nunbers I have here are before tax transfers and we are quite bad in that regard.

    https://tradingeconomics.com/country-list/gini-index-wb-data.html

    ReplyDelete
    Replies
    1. I've already verified with Singstat and UK's Office of National Statistics. They both provide after tax & after transfers Gini.

      I use sites like Tradingeconomics and Statista for quick and dirty lookups. But for real verification, it's better to go direct to the primary sources.

      Delete
  3. Too cheem (profound). I just treat the CPF as part of the 40% less risky portion in my approx 60/40 portfolio.

    ReplyDelete
    Replies
    1. That is actually more realistic than the report.

      Delete