This dissertation aims to promote our understanding of financial markets, particularly focusingon over-the-counter (OTC) markets. Contrary to exchanges, where all participants can trade at the same prices, in OTC markets, trading is bilateral. On exchanges trading is anonymous, so participants do not know with whom they are trading. In OTC markets bilateral trading implies that each participant knows with whom they are currently trading. Therefore, intermediaries, called dealers or liquidity providers, provide personalized prices to each participant. Understanding the provision of personalized prices is an important area of research. In the past, data limitations provided obstacles for this area of research. Before the introduction of Trade Reporting and Compliance Engine (TRACE) in 2002, data in OTC markets were generally not available. TRACE collects and timely disseminates all transactions involving dealers in the corporate bond market, making information on corporate bond prices and traded quantities available to all participants, not just the participants of the trade. While TRACE is specific to the US corporate bond market, similar Reporting Facilities emerged in the municipal bond market. Once introduced, TRACE revealed that transaction costs, the difference between the price paid and the value of the corporate bond, were high and very variable across clients. Over time, TRACE has revealed additional information, most notably a unique identifier for each dealer. This identifier allows us to study how prices, and thus transaction costs, vary across different dealers. To this day, TRACE or the corresponding repositories of all trades in other markets, like the municipal bond market, are common data for analyzing OTC markets. The data includes all trades and identifies the dealers in the transaction. However, many OTC markets do not have a repository collecting all trades, most notably the foreign exchange (FX) markets, an OTC market, and the world’s largest financial market. Also, the existing repositories have multiple drawbacks. 2 Firstly, repositories do not reveal the identity of a client in a trade. This lack of identification means that only outcomes for the average client of a dealer can be studied. One cannot observe differences across clients. Thus differences across dealers for different clients cannot be determined. Furthermore, from a client’s point of view, differences across dealers are only comparable if the client is the average client for each dealer, i.e., the average client across dealers is the same. Secondly, repositories do not observe trade initiation. So it cannot be determined which participant contacted the other participant asking to trade. Instead, the literature commonly assumes that clients always initiate trades when trading with dealers. However, for interdealer trades, trade initiation cannot be assigned to either dealer in this way. Furthermore, in riskless principal transactions, where a dealer trades with two clients simultaneously, only one client initiated the trade with the dealer, while the dealer likely initiated the trade with the other client. However, which of the two clients initiated the trade can also not be assigned. Thirdly and finally, the repositories only collect information on trades. Many participants have multiple dealers from whom they elicit quotes when wanting to trade. Repositories do not observe any of the quotes that dealers provided but did not lead to trade. However, quotes reveal important information about trading and the market functioning. Quotes reveal which dealers clients contact to trade and whether these dealers respond. In addition, quotes allow to measure competition in trades. More generally, quotes allow to observe in much greater detail why transaction cost differ across clients. To counter these shortcomings, research has to rely on alternative data. This dissertation utilizes data from a leading trading platform in the FX market that addresses these shortcomings. The Bank of International Settlement’s "Triennial Central Bank Survey of Foreign Exchange and Over-the-counter (OTC) Derivatives Markets in 2019" mentions six multi-bank platforms in the FX market. One of those six platforms provides the data for this dissertation. The data allows to identify both dealers and clients and provides information about the trade initiator, i.e., the liquidity demander. Most importantly, however, the data observes the quotes that liquidity demanders receive from all their dealers. To my knowledge, it is the first research to observe and utilize these three pieces of information. In the first chapter Chapter 1: Client-Dealer Intermediation in OTC Markets, I utilize said data to analyze the prices that small clients receive in OTC markets and how they can improve their prices. I show that clients realize better prices when contacting more dealers, and they receive better prices from a particular dealer when trading more with the dealer. Small clients can only satisfy one of these conditions, as trading with many dealers means trading only little with each dealer. Utilizing the quote data, I find that clients have a third option. They can trade with a type of dealer I call "match maker." Upon being contacted by a client, a match maker immediately contacts other dealers, receiving quotes from those dealers and relaying the best 3 quote she receives to her client at a mark-up. The match maker then trades with the dealers if and only if the client trades with her. Only alternative data, in this case, quote data, can unveil this type of dealer, as I need to observe the behavior of these dealers when they do not trade. Match makers effectively pool their clients trading together under one identity and thus overcome the disadvantages of being small. In doing so, match makers enlargen their client’s networks of dealers by at least 40% and reduce the trading cost for their small clients by 20%. I show that match makers are prevalent, with 14% of dealers being match makers and 12% of clients trading with them. In the second chapter Chapter 2: Centrality in OTC Markets, Liquidity Provision, and Prices, I study the question of how the choice and number of dealers affect the prices that clients receive. This choice is a much-studied question in many different OTC markets, generally finding that more central participants profit when trading, i.e., a centrality premium exists. Central participants are participants with many trading partners, while peripheral participants have few trading partners. This research is the first to observe trade initiation, i.e., liquidity provision. As in the previous literature, without conditioning on liquidity provision, I find a centrality premium, i.e., the more central participant makes a profit in the trade. However, the more central participant generally provides liquidity in a trade. Once I control for liquidity provision, I find that all liquidity demanders, independent of their centrality, pay a spread for demanding liquidity. Furthermore, this spread is larger the less central the liquidity demander is. The centrality of the liquidity demander mainly determines the size of the spread, the centrality of the liquidity provider is less important. Central liquidity demanders receive quotes from many liquidity providers simultaneously. For them, I find that more central participants trade more often with them. At the same time, the trade prices are similar across the centrality of the liquidity provider. Peripheral liquidity demanders only contact one or a few liquidity providers. For them, I find that they trade at better prices when trading with more central liquidity providers. Taken together, I infer for all liquidity demanders that the more central a liquidity provider is, the better the average price that the liquidity provider quotes. Thus, I find a centrality discount once controlling for trade initialization. The centrality premium observed in previous research is due to liquidity provision by more central participants. Conditioning on the liquidity provision in a trade, I find a centrality discount. The first two chapters of this dissertation study how transaction costs differ across clients in the FX market, with a focus on especially small clients. In particular, the first chapter shows how small clients can reduce their trading costs. In both chapters small clients are small corporations or small institutional investors, still trading at least hundred of thousands a month. Truly 4 small investors are households. Compared to institutional investors, many households find it difficult to access financial markets. The final chapter explores whether a particular form of financial innovation, namely Robo-Advisers, help making financial markets more accessible to households. In the final chapter Chapter 3: Robo-Advisers: Household Stock Market Participation and Investment Behavior, I study whether households benefit from financial innovation. Like small clients in the OTC market, households find it difficult to participate in financial markets. However, instead of facing high monetary costs, many households lack the proper education, i.e., financial literacy, to be willing to participate in financial markets. Financial literacy is generally low and difficult to improve. In this work, I study whether financial innovation, in the form of a Robo-Adviser, encourages households to invest in financial markets. This work is the first to study a fully automated investment process that offers advice to households with much lower net worth than human financial advisers require. I find that one-third of all Robo- Adviser users are new to financial markets, significantly higher than for regular retail investors. Furthermore, these households would not have invested in financial markets were it not for the introduction of the Robo-Adviser. I conclude that Robo-Advisers are an important tool to encourage participation in financial markets, providing households with academically-vetted portfolios.