The fallacy of foreign flows and how South Africans are ‘fooled by randomness’ – Rowan Williams-Short

Human beings, particularly in this modern age, have a tendency to view the world as more explainable than it is and very often look for explanations where, in fact, there simply are none. This tendency serves us as, in the absence of explanations, we are confronted with randomness, something most people find unacceptably uncomfortable. Foreign flows – Foreign Direct Investment (FDI) – record the value of cross-border transactions related to direct investment during a given period of time, usually a quarter or a year. This article by Rowan Williams-Short, the Head of Fixed Interest at Vunani Fund Managers, poses the idea that South Africans’ unusual fascination with foreign purchases and sales of domestic equities and bonds and, in particular, their belief that data in respect thereof is essential to formulating predictions is nonsensical. Williams-Short reasons these data points are utterly unusable in forming any reliable anticipations as they are purely random. The premise of Williams-Short’s argument brings to mind the thesis presented by Nassim Nicholas Taleb in his book Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets. In the book, Taleb deals with the fallibility of human knowledge and sets forth the idea that modern humans are often unaware of the existence of randomness. – Nadya Swart

The fallacy of foreign flows (FFF?)

By Rowan Williams-Short

South Africans have a peculiar and unusual fascination, bordering on an obsession, with foreign purchases and sales of domestic equities and bonds. Some prominent economists and strategists have described these data points as essential and critical to forming their forward-looking views of asset classes. 

I submit:

  • Foreign purchases of South African equities or bonds have no bearing whatsoever on the prices of those securities, nor should they.
  • The time series data of those net purchases reveal them to be so random – the physics equivalent of white noise – that they are utterly unusable in forming any reliable anticipations.

As a buy-side portfolio manager, I receive at least five emails a day from banks and brokers who have recorded the previous day’s net foreign purchases. These are sent on nothing more than misplaced faith that they are useful for understanding the previous day’s market moves (as though that ought to matter to investors anyway). In a dreadful display of the behavioural finance affliction known as ‘confirmatory bias’ (or to my mind, academic dishonesty) the data is treated in one of two ways. On those days that the market moved in the direction consistent with a belief that foreign flows influence prices, the data is enthusiastically seized upon as being explanatory. On the roughly equal number of days on which the opposite has occurred, the silence is deafening. 

Initially, as a heuristic or an instinctive reaction, one can understand and even sympathise with this trap. Global markets are axiomatically much larger than the South African market (and every other single country market, for that matter). So, if a potentially large amount of money enters or exits, that must surely move prices, right?

The problems with that belief are manyfold. I shall mention some.

i. Financial markets are sophisticated. The crassly simplified economics of supply and demand that might work in rudimentary bartering of bread and butter do not apply. Two extreme examples spring to mind. First, when erstwhile domestic market darling Dimension Data in its halcyon days not only listed in London, but was fast-tracked into the FTSE100, any number of otherwise contemplative analysts leapt to the false conclusion that ‘forced buyers’, including slavish index trackers, must push up the price. On listing day, there was a spike at the close that lasted for a few minutes (so hardly anybody participated and spare a thought for the presumably passive latecomer who was on the buy leg). The next morning the price was right back to its listing level. Then DiData began its ignominious 98% plunge. I note in passing that DiData was on a Panglossian PE multiple of 72 when it listed. Second, on both entry to and exit from the WGBI (World Government Bond Index) South African bonds were widely deemed to be in for a bull and bear market respectively. In fact, neither transpired, as can be easily demonstrated by anyone who takes the trouble to account for conflating factors at both times.

ii. After a moment’s reflection, why should the geographic location of the buyer or seller impinge upon prices? Before foreigners became meaningful participants in South African markets (i.e. before 1994) asset managers in Cape Town did not speculate on whether Johannesburg market participants might turn buyers or sellers on a given day, week or month. A buyer or seller, by dint of important regulations, is not privy to the identity, far less location, of whomever is on the other side of his or her trade. It matters not a jot whether it transpires that the counterparty to my trades is based in London or in East London. Listed equities and bonds are traded anonymously and are fully fungible.

iii. Focusing solely on net foreign purchases is a lazy approach to research. When one reads, for example, that “Yesterday foreigners were net sellers of R1 billion of South African bonds” one is well advised to disaggregate purchases and sales. On a typical day with that volume recorded, it will transpire that in fact some foreigners bought R21 billion while other foreigners sold R20 billion. Very often, both the buyer and the seller in any transaction are both foreigners, unbeknown to one another. Don’t conjure up an image of all foreigners meeting up for coffee on Sunday nights, colluding on their South African trades on Monday. Foreigners are not ubiquitous creatures of one mind.

iv. To believe that foreigners consistently move prices in a predictable direction, you’d need to believe that they were less price sensitive (relatively price elastic in economic parlance) than South Africans. That is self-evidently a preposterous proposition.

v. South Africans still account for larger volumes on the JSE than do foreigners.

vi. Finally, for heaven’s sake, let’s check the facts, which turn out to be irrefutable. I have calculated the R-squared measures between both the level of the All Share Index and net foreign equity purchases, and the levels of the constant maturity 10-year South African government bond yield and net foreign bond purchases. Using data from 1994 to October 2021, these are 1.4% and 5.6% respectively. For a convenient point of reference, the correlations are considerably lower than the correlation found between the seasonal arrival of storks in Europe and the births of human babies, but we don’t hear analysts deducing that babies come from storks (*).

It is well known that [high] correlations do not prove causality. Less frequently cited is the corollary that low correlations rule out causality.

Investors have more than enough dynamic complexities with which to grapple. The good news is that you might as well expunge foreign flows from your job jar; following those is futile.

Postscript: I am aware that there has been some doubt about recent data releases. For the great majority of the period that I have researched, JSE data was sufficiently accordant with other sources of data. In any case, the a priori reasoning I have given ought to be conclusive.

(*) The low but interesting correlation between storks and babies is easily explained: more humans = more houses = more chimneys = more roosting sites for storks.

  • Rowan Williams-Short is the Head of Fixed Interest at Vunani Fund Managers. 

Read also:

Visited 2,780 times, 2 visit(s) today