Covid,yieldsandchinksinthemodel

Statistics is generally cold and often mundane as it often fails to convey the story behind those numbers but covid stats in India are no longer cold they are chilling. The new case numbers are rising daily, and it reached an all-time high of 261k on Saturday. The daily deaths number is also inching up daily. Compare this to the world statistics where countries like the US are showing a much-reduced daily incidence number even when the economic activity is reopening. Compare this to a situation of around mid-February where numbers were in the range of 12k-15k daily and the corona war was supposedly won already and congratulations were already exchanged. Now the models have suggested that numbers would continue to peak for at least one month more before plateauing. We will discuss the models in detail below.

In the global news, the Dow Jones continued its bull run ending at all-time highs on Friday. The Hang Seng and Nikkei are also trading in green in the morning trade today. The good data on the economy and inflation being in check presents the perfect recipe for the stock markets to do well. US yields are down, the 10-year is trading at around 1.56 levels which is a significant climb down from 1.75 seen on March 31^st. Close to 175 bn USD bond auctions in the previous week didn’t result in a yield spike as markets absorbed the new issuances smoothly. On the other hand, a lot of action happened in India GSec auctions on Friday where the auction of 10-year paper was cancelled. The underwriting fee of the paper had reached the highs of 47 paise. The cancellation resulted in a lot of short covering and the yields coming down to close at 6.09 against the pre-auction level of 6.15. Rising yields are taking into account the uncertainty in government finances post the second spike.

Now let’s come back to our discussion of models which we referred to earlier regarding the forecasted covid trajectory. How competent can these assessments be? The question is important because a lot of many other forecasts like GDP growth, inflation and resource allocation in the economy will be dependent on the starting premise of how we see covid panning out. In their book Radical Uncertainty: Decision making for an unknowable future, author duo John Kay and Mervyn King lament about the dependency on models. They write that modern economists are always in search of developing ever more complex models to represent reality and forecast the future (ECB uses a model called DSGE ie Dynamic Stochastic General Equilibrium model to make its GDP assessments). As per the authors these models which depend heavily on past data sometimes prove futile and sometimes even dangerous when they are made to predict a phenomenon plagued with radical uncertainty. Authors write that the basic fault lies on two points, one is the assumption of stationarity i.e. the processes underlying the model remains stationary. In the case of covid the behaviour of the public changes as the numbers grow, government laws become stricter or vaccine rollout increases. To mathematically capture this reflexive behaviour is difficult to say the least. The second factor is about the shocks exogenous to models which are due to unforeseen radical changes. A super-spreader event here, a chance discovery of vaccine there can alter the end result in a substantial manner. The authors finally add that the models often break down when the faith in them is the highest. The example which they give is of the models pricing the CLO’s and MBS’s just prior to GFC. One of the authors who was the Governor of the BOE for 10 years writes that he was quite embarrassed when the Queen of England asked him why you didn’t see the GFC coming. He wasn’t able to resort to model bashing at that time.

Concluding the note today, we will say that in the highly uncertain times like which we are living in being conservative is preferable. Safety of models howsoever sophisticated they are is questionable.