The Model Thinker #16 : Markov Models

Markov models capture systems that transition probabilistically between a finite set of states.

One application in finance are Bond credit ratings. A "AAA" bond has a 95% every year of retaining it's high credit rating but there is a 5% of it being downgraded to "AA". In similar vein a "CCC" bond has a probability to be upgraded to "B" or go into default. Credit analysts construct a table of transition probabilities to gain a helicopter view of the bond markets.

A Markov process can converge into a unique statistical equilibrium if it satisfies four conditions :

a) It has a finite set of states.
b) Probabilities of transition between each state is fixed.
c) The system can get from any state or any other state through a series of transitions.
d) The system does not produce a deterministic cycle through a sequence of states.

If only these four conditions are met, the number of bonds of a particular credit rating would converge to a fixed percentage over time. Unfortunately, transition probabilities do not remain constant over different part of market cycles. The probability of a bond defaulting goes up in a bear market.

There are a some lessons we can learn from Markov models.

For history to matter, transition probabilities must change.

A one-shot intervention in a food drive creates only a temporary boost and cannot resolve the problem of inequality. As such, beyond virtue signalling, Professor Teo You Yenn's self sacrifice to deny her children her own social advantages does jack-shit for fellow Singaporeans. Instead, pushing for a serious program like the Professional Conversion Program would create a lasting effect because, if done properly, it can increase the probability of transitioning unemployment to being employed. I suppose policy tweaks need to make this work for non-degree holders.

Imagine you can either be employed or unemployed. Career management would mean reducing the probability of transitioning from employment to unemployment. This may mean playing office politics or putting yourself under a 996 work regime to avoid getting retrenched. This can be tiring.

The other possibility is to increase your probability of becoming employed if you fall into a state of unemployment. This is where retraining comes in, but networking and helping folks in your weak networks matter as well because you can call in a favour later in your career.

Dividends investing belong to a totally different class of intervention strategies that makes unemployment totally irrelevant so it can complement your career management strategy moving forward.

Unfortunately, to succeed in dividends investing you may need to manage your career better in the short term so as to generate the funds to build your portfolio in the first place.

There are no short cuts in life.