This thesis contains three papers on market microstructure. The papers study the measurement of trading costs, the enforcement of insider trading regulation, and corporate insiders’ choice of trading venue. A common thread is how information is reflected in market data and how that data is used by researchers, regulators, and market participants.

Article I (with Björn Hagströmer) studies bias in the effective bid–ask spread. Standard algorithms assume that the relevant quote always precedes the trade by a fixed positive lag. We show that relative trade latency is stochastic and can be negative. Using data from the London Stock Exchange, we find that the Lee–Ready algorithm overstates the midpoint-based effective spread by 8.4% and the depth-weighted effective spread by about 18%. We introduce a new matching algorithm that eliminates the bias.

Article II studies the full enforcement chain for illegal insider trading, from suspicion to prosecution. Using proprietary data on all suspicious transaction and order reports filed with the Swedish Financial Supervisory Authority between 2016 and 2019, I model enforcement as a three-step process. Only 2.12% of suspected individuals are prosecuted. Market-based signals drive the initial detection step. The suspect’s connection to the firm and the value of the information determine forwarding and prosecution.

Article III (with Lars L. Nordén) examines where corporate insiders trade when the same stock is available on exchanges and dark markets. Using data on Swedish insider transactions from 2016 to 2024, we find that when insiders are buying, they are more likely to trade on dark markets when engaging in illegal activities, but less inclined to do so when they are informed. Given insiders’ endogenous venue selection, buying on dark markets negatively impacts abnormal returns. When insiders are selling, their venue choice is unrelated to whether they are informed or engage in illegal activities, and trading on dark markets does not significantly affect abnormal returns.