The Effect of Algorithmic Trading on Voluntary Disclosure

Andrew P. Stephan

doi:https://doi.org/10.21985/N2S469

Work

The Effect of Algorithmic Trading on Voluntary Disclosure

Public Deposited

Download PDF

Download All Files (.zip)

Algorithmic trading (AT) has grown dramatically in recent years and now makes up over half of all trades and orders in the market. I investigate whether and how AT affects voluntary disclosure by managers. I hypothesize that AT’s differential ability to process information and its speed of trading affects how closely the market price aligns with the firm’s fundamental value, and thereby changes the equilibrium disclosure level for managers. First, I investigate whether AT affects the quantity of voluntary disclosure from managers. Second, I examine why AT has such an effect. Last, I explore whether AT affects other disclosure characteristics such as specificity, readability, and timeliness. This question is important to academics, regulators, and the investing public due to the current debate over the desirability of algorithmic and high frequency trading in capital markets. In Chapter 2, I define algorithmic trading, discuss its role in the capital markets, and review the prior literature examining its effects on price informativeness. AT utilizes many diverse strategies in the market, including market making, trading on statistical arbitrage opportunities, trading quickly in response to news, and executing large orders at efficient prices. The prior literature finds AT prices known information more efficiently, but it may reduce the amount of information known, leading to a net decrease in informativeness. In this chapter, I also review the literature regarding managers’ forecast decisions. In Chapter 3, I investigate the effect of AT on the quantity of voluntary disclosures, as proxied by management forecasts. I find the level of AT leading up to an earnings announcement is positively associated with the likelihood of issuing at least one forecast, the quantity of forecasts issued, and the number of days on which the firm issues forecasts. I utilize the introduction of the NYSE Autoquote, which exogenously increased AT for NYSE firms in 2003, as a quasi-natural experiment. I find results consistent with my main test, namely that NYSE firms see a greater increase in forecasts post-treatment compared to a matched sample of NASDAQ and AMEX control firms. In Chapter 4, I examine why I find a positive association between AT and disclosure. Prior literature suggests that AT may decrease price informativeness by reducing the amount of information acquired by investors (Weller 2016). Under this explanation, managers would disclose more to offset this reduced information acquisition. I perform three tests to determine whether AT reduces the amount of information acquired by investors. First, I examine management forecast response coefficients, finding returns are more strongly associated with forecast surprises when AT leading up to the forecast issue date is higher. This is consistent with the market being less informed when the management forecast is released. Second, I test information acquisition directly, finding AT is negatively correlated with both EDGAR downloads and the magnitude of analyst forecast revisions. Moreover, AT is positively associated with the cost of informed trading, which may explain the decrease in information acquisition. Last, I examine whether the type of AT that improves price informativeness has the opposite effect on disclosure, and find that it does. In Chapter 5, I investigate the association between AT and other disclosure characteristics. I find mixed results regarding AT and the specificity of disclosures. AT is negatively associated with forecast specificity, where point forecasts are the most specific, closed range forecasts are next, and open range forecasts are the least specific. For closed range forecasts, however, AT is negatively associated with the absolute range of the forecast, indicating these forecasts are more specific. I find no association between AT and the timeliness of disclosures and the readability of disclosures.

Last modified