Are investment models only an exercise in putting false certainty around an inherently uncertain world? asks David Jane, fund manager, Premier Miton Macro Thematic Multi Asset Team, Premier Miton Investors.
I have always enjoyed maths; I like the precision and the certainty it provides. At university I chose to study subjects such as maths, statistics, econometrics and financial mathematics alongside economics. Going into the City after university I was sure that these skills would provide some advantage. Particularly, as around that time computers were moving from being academic tools to being available on office workers’ desktops, we could now build complicated models of the financial world.
Over time reality has shattered those naïve early expectations. One of my first experiences in the fund management industry was the aftermath of the 1987 crash, where people were questioning why the portfolio insurance models actually led to the crash they were meant to protect against.
A few years later we had the rise and fall of LTCM. The ‘brightest finance PhDs’, including two Nobel prize winners, had been gathered together to run a hugely leveraged hedge fund based on really complex financial models that showed they were taking minimal risk. Once something occurred outside their models, in this case the Russian financial crisis, the firm went rapidly bust, nearly taking the entirety of Wall Street down with it.
I was working at the time as an analyst, building beautiful and complex forecasts and valuation models for the stocks I was following. While satisfying, all this energy being expended was an exercise in creating false confidence and certainty, as was starkly highlighted by the ‘Dot Com bubble and crash’. Analysts were projecting confidently the future for companies and their models were providing the evidence for huge valuations. Ultimately reality trumped the models, and we had the dot com crash. One prominent victim of this was Enron, whose use of ‘mark to model’ accounting was one factor in its demise.
A period of relative stability prevailed in the aftermath of the ‘dot com’ crash, but this period was one during which a large amount of financial innovation was taking place. Mortgage markets had been deregulated in most jurisdictions and these newly created mortgages were being packaged into collateralised mortgage obligations, which themselves were then sliced up into tranches and sold off as very low risk investments. Even the riskiest mortgages, which involved no income appraisal (known even at the time as ‘liar loans’), could be packaged up as low risk assets on the basis that the risk of a diverse portfolio of such loans was less risky than any individual loan. Banks were being encouraged to make loans into less privileged markets by regulation, so the stage was set for a housing bubble and subsequent crash.
The same was true of CLOs and CDOs (collateralised loan obligation and collateralised debt obligations), poor credits could be added together into pools to turn the result into a strong credit, the maths proved it. Sadly, the maths ignored the fact that they might all go wrong at the same time, due to the existence of economic and market cycles. I saw first-hand how a supposedly low risk asset could almost instantly become worthless because something occurred that the model didn’t allow for.
Much more recently we had the UK pension fund crisis, where the pension funds had been borrowing against their, supposedly, low risk long dated government bonds, in order to invest in more attractive ‘higher return’ assets. A great way of enhancing returns, whilst still meeting their obligation to hold long duration assets to meet their regulatory requirement, again, the models proved it! This all worked very well until the spike up in bond yields worldwide and the lenders called in their collateral. This led to forced selling of UK government bonds and the Bank of England having to step in to prevent a rout.
Why am I writing about these historical events now? Financial modelling is clearly a feature of fund management and the investment advisory business. In many cases we are encouraged for regulatory reasons down that path, in others it’s an exercise in putting false certainty around an inherently uncertain world. It’s always highly tempting to find an historical relationship, build a complex model around it and tell clients that you’ve found the secret sauce. Which works until it doesn’t.
In my view it’s a simple fact that it’s the models that create the circumstances for their own failure. Once it becomes ‘common knowledge’ that there is an approach that works, the pricing that previously prevailed is disturbed. Consider the low quality loans that backed CLOs. The creation of these instruments with their high-quality credit ratings, created artificial demand from investment grade buyers and therefore drove up pricing such that the supposed anomaly no longer existed. The same with most of the other opportunities that turned disastrous. This applies to the 60/40 approach to asset allocation, expecting bonds to provide good diversification for equities when the starting yield was sub 50bps was clearly irrational, but the models said it would work.
My conclusion is to stick to a common-sense approach, if it seems too good to be true then it surely isn’t. Stick with the hard known facts of today, perhaps the starting yields for a post-retirement income strategy, or current pricing for a total return strategy and use this as the basis for decision making, rather than some clever financial mathematics cooked up to confuse.
Main image: ux-indonesia-8mikJ83LmSQ-unsplash
