Recently by chance I came across a paper (see Reference section) titled The boats that did not sail: Asset price volatility in a natural experiment. It grabbed my attention because it proposed a very creative way of applying approaches in natural science to the field of finance. This paper studied capital markets with very long information delay: the London and Amsterdam exchanges back in the 18th century, which at the time were 3-4 days away from each other. Having learned a few things from this paper, I share my understanding and reflections of it in this article.
The Koudijs paper
This paper discussed the impact of three factors in asset pricing (more specifically, short term price volatility): public information, private information and liquidity shocks. Public information are information disclosed through official news channels and generally available to everyone; private information refers to information not publicly disclosed, only accessible to a small group of investors; and liquidity shocks are price variance in an asset caused simply by trading this asset. When trying to quantify the impact of each factor, a 21st centry citizen would immediately see the difficulty: in the era of information boom, it is very difficult, if possible at all, to dissect public information away from private ones. Private information may be made public in the matter of miliseconds after they are born. Anyone with a smartphone can get the latest push information from any market research firm. How can we find a situation where public information no longer an actor on the stage?
The author found his answer in history. He studied the financial markets in London and Amsterdam in the 18th century, a time where information was still passed on by packet boats instead of internet packets. Since some British stocks were listed in both London and Amsterdam, information about them had to travel from London to Amsterdam, by mail. The information channel was unique, slow and most importantly, subject to exogenous conditions, namely the weather. When the weather doesn’t permit sailing across the North Sea, no information about the British companies can get from London to Amsterdam. That is simply the perfect scenario where public and private information don’t mix and can be put under the microscope separately.
With this brilliant approach, the author observed some interesting phenomena: when sailing conditions are good, British stock prices in Amsterdam seemed to follow the general trend of the London ones, with a few days of delay and some additional variance; this is somewhat expected. The surprise happens when sailing (and therefore, information) is blocked by weather: without information flow, London and Amsterdam exchanges are two separate worlds, and yet, British stock prices were moving very consistently with each other! The obvious corollary is, other than public news, there are other big sharks swimming deep in the ocean of asset pricing. Having found a way to isolate public information into its own silo, Koudijs eventually arrived at the conclusion that public info, private info and (transitory) liquidity shocks account for about 40%, 30% and 30% of overall price variance, respectively.
My thoughts and reflections
I appreciate this paper so much as to share it on my blog, not only because it is a hybrid of natural and social science, but also it demonstrates how to think outside the box: when ordinary databases didn’t have data sets readily available, data sets in which public information can be ignored in the market, the author recognized the reason for such a difficulty in today’s finance study, and turned to other subjects (science and history) for more insights.
Taking it one step further, in an almost sci-fi sense, this paper may even shine some lights for anyone interested in studying the future: imagine, if someday we the humans colonized the other solar planets, well enough that the NYSE has a Martian branch/counterpart, we may very well face the situation discussed in Koudijs’ paper again. There, information flow is naturally slow, bounded by the speed of light, and can take a long time to get from planet earth to mars. There, this paper may cast some insights in how the capital asset pricing might work again.
As an example, we may take Koudijs’ argument on arbitrage, that (in this London-Amsterdam setting) perfect arbitrage cannot exist because of the time effect of arbitrage is critical, yet it would take at least 7 days for an arbitrage operation to complete. Then by this logic, consider the earth-mars setting, where the information flow is only less than an hour for a round trip for communication signal (traveling at speed of light); it maybe reasonable to assume that arbitrage would be much closer to perfect in this setting.
Another example: we see from this paper that, just by existing, private information can impact the market, even if they were never meant to do so. Then we may ask: can private information stop existing? Could our society eventually evolve into one where truly private information simply doesn’t exist anymore? Information is truly private if only known to creator and designated receiver. But even nowadays, information we make have its (many) third-party observers: when we send an email, it’s read by spam filters (a program) before by the recipient; when we share location, our map app’s user experience program (a program) knows that location before our friends do; not to mention communication surveillance performed by government agencies (humans). It isn’t crazy to imagine such third-party information access can grow into a saturated state. Then, there, even if none of these third-parties mean to throw these information at the financial market directly, such information stop being completely private. What would that mean, if we believe in the significance of private info in asset pricing? Answers to such questions can potentially be inspired from works similar to this paper of Koudijs’.
Reference
Koudijs, Peter. “The boats that did not sail: Asset price volatility in a natural experiment.” The Journal of Finance 71.3 (2016): 1185-1226.