Every Cloud Has a Silver Lining: Fast Trading, Microwave Connectivity, and Trading Costs

• Author: KONSTANTIN SOKOLOV,ANDRIY SHKILKO
• Published date: 01 December 2020
• Date: 01 December 2020
• DOI: http://doi.org/10.1111/jofi.12969

THE JOURNAL OF FINANCE •VOL. LXXV, NO. 6 •DECEMBER 2020

Every Cloud Has a Silver Lining: Fast Trading,

Microwave Connectivity, and Trading Costs

ANDRIY SHKILKO and KONSTANTIN SOKOLOV∗

ABSTRACT

Modern markets are characterized by speed differentials, with some traders being

fractions of a second faster than others. Theoretical models suggest that such differ-

entials may have both positive and negative effects on liquidity and gains from trade.

We examine these effects by studying a series of exogenous weather episodes that

temporarily remove the speed advantages of the fastest traders by disrupting their

microwave networks. The disruptions are associated with lower adverse selection

and lower trading costs. In additional analysis, we show that the long-term removal

of speed differentials results in similar effects and also increases gains from trade.

COMPETITION ON RELATIVE SPEED IS A DEFINING characteristic of modern

markets where trading firms invest heavily to gain a speed advantage over

their rivals. The race to acquire the fastest technology often leads to subsec-

ond speed differentials among traders. A rich theoretical literature suggests

that such differentials may have opposing effects on liquidity and gains from

∗Andriy Shkilko is with Lazaridis School of Business, Wilfrid Laurier University. Konstantin

Sokolov is with Fogelman College of Business and Economics, University of Memphis. We thank

Stefan Nagel (the Editor); an anonymous Associate Editor; two anonymous referees; Jim Angel;

Matt Barron; Robert Battalio; Ekkehart Boehmer; Jonathan Brogaard; Michael Brolley; Eric Bud-

ish; Adam Clark-Joseph; Jean-Edouard Colliard; Amy Edwards; Sean Foley; Thierry Foucault;

Michael Goldstein; Björn Hagströmer; Terrence Hendershott; Peter Hoffmann; Albert Menkveld;

Sophie Moinas; Peter O’Neill; Andreas Park; Lasse Pedersen; Fabricio Perez; Richard Philip; Bar-

bara Rindi; Ryan Riordan; Elvira Sojli; Eric Stockland; Wing-Wah Tham; Erik Theissen; Tugkan

Tuzun; Brian Weller; Jonathan Witmer; Bart Yueshen; Haoxiang Zhu; and the audiences at the

Central Bank Microstructure Workshop, Bank of Canada, CFTC, Conference on the Econometrics

of Financial Markets, EFA,FCA, Finance Down Under, FIRS, NBER Market Microstructure Meet-

ing, Northern Finance Association, Paris Finance Meeting, SGF, Chapman University, University

of Memphis, University of Sydney, and the WFE-Imperial College London Conference for com-

ments. Daniel Ganev and Jiacheng Zhou provided research assistance. Stuart Hinson from NOAA

and Jireh Ray from the CME provided data guidance. We acknowledge financial support from the

Canada Research Chairs program, Canada Foundation for Innovation, Ontario Early Researcher

and Graduate Scholarship programs, and the Social Sciences and Humanities Research Council

of Canada. No party had the right to review the paper prior to its circulation. We have read The

Journal of Finance’s disclosure policy and have no conflicts of interest to disclose.

Correspondence: Andriy Shkilko, Lazaridis School of Business, Wilfrid Laurier University;

e-mail: ashkilko@wlu.ca.

DOI: 10.1111/jofi.12969

© 2020 the American Finance Association

2899

2900 The Journal of Finance®

trade.1On the one hand, speed may allow liquidity providers to reduce their

adverse selection exposure and manage inventories more efficiently. Alterna-

tively,speed may allow traders to pick off limit orders before liquidity providers

adjust to new information. The former effect has a positive impact on liquidity

and may increase gains from trade, whereas the latter may have the opposite

impact. To shed light on the net effect, we examine a two-year time series of

exogenous weather-related intraday shocks to speed differentials. The results

indicate that when the differentials exist, liquidity is impaired. In an addi-

tional analysis, we show that the long-term removal of speed differentials is

associated with liquidity improvements and greater gains from trade.

In the main analysis, we examine liquidity when heavy precipitation dis-

rupts microwave transmissions between Chicago and New York. During our

2011 through 2012 sample period, traders send information between the two

cities via either a fiber optic cable or a microwave network. The microwave net-

works, which are about 30% faster than the cable, have two important char-

acteristics. First, only a small group of trading firms has access to them, and

these firms engage in constant competition for the top speed by retrofitting con-

tinuously. Second, precipitation (i.e., rain and snow) disrupts them. The first

characteristic creates a speed advantage for select traders, whereas the second

characteristic occasionally removes this advantage. We show that when the

microwave speed advantage is removed, adverse selection and trading costs

decline by up to 6.7% and 5.2%, respectively.

The effect of precipitation on microwave communications is well known in

physics, but has not previously been examined in the financial markets con-

text. To confirm that precipitation does indeed serve as a shock to information

transmission speeds, we show that equities in New York react to futures sig-

nals from Chicago two milliseconds slower when it rains or snows. During our

sample period, two milliseconds is precisely the difference between the mi-

crowave and fiber optic transmission speeds. Furthermore, the data show that

when microwave signals arrive from the futures market, equity arbitrage al-

gorithms pick off the stale quotes, while liquidity-providing algorithms reprice

their quotes. As such, consistent with the theoretical literature, market partic-

ipants rely on speed to both supply and demand liquidity.

Given that speed is used for both liquidity supply and demand, how do the

abovementioned adverse selection costs emerge? To clarify the mechanism, it

is useful to think of two groups of market participants: Group 1—those with

access to microwave networks, and Group 2—those without such access. Partic-

ipants from both groups often act as liquidity providers. As such, the National

Best Bid and Offer (NBBO) is a combination of quotes from many competitors.

When futures reveal information that is not yet reflected in equity prices, and

1See Hoffmann (2014), Biais, Foucault, and Moinas (2015), Budish, Cramton, and Shim (2015),

Foucault, Hombert, and Ro¸su (2016), Jovanovic and Menkveld (2016), Aït-Sahalia and Saˇ

glam

(2017), Foucault, Kozhan, and Tham (2017), Menkveld and Zoican (2017), and Baldauf and Mollner

(2020), among others. We review this literature and relate our findings to its predictions later in

the Introduction.