کشف قیمت در بازارهای اوراق قرضه دولتی

Price discovery in government bond markets is explored using Norwegian data including trades from both tiers of the market and dealer identities. The results show that while aggregate interdealer order flow explains one-fourth of daily yield changes, aggregate customer order flow has little explanatory power. Dealers are heterogeneously informed and appear to have different sources of information. While some dealers mainly rely on their customer trades, others appear to rely on skill in acquiring and interpreting other relevant information, suggesting that dealers play an independent role in the price discovery process.

The process of price formation is largely ignored in traditional economic models that assume that prices adjust instantaneously to new information. This assumption applies equally to public and private information. The market microstructure literature advocates a different view. In microstructure models, the process of price formation is crucial and private information plays a key role. Information is incorporated into prices through two channels: a direct channel where prices adjust immediately to new public information, and an indirect channel, referred to as price discovery, where prices adjust over time to private information conveyed through order flow.1 Lyons (2001) defines private information as information, not known by all, that produces a better price forecast than public information alone. Several studies confirm that price discovery is an important part of the price formation process in asset markets. Hasbrouck (1991), Evans and Lyons (2002), and Brandt and Kavajecz (2004) document that contemporaneous order flow explains a substantial part of daily price changes in stock markets, foreign exchange markets, and government bond markets, respectively. The gradual adjustment of prices through order flow implies that prices do not fully reflect all available information at any point in time and suggests that market participants have asymmetric information. The purpose of this paper is to investigate the role of different market participants in the process of price discovery in government bond markets. Government bond markets are characterized by a two-tier structure, reflecting that customers can trade with dealers only while dealers can trade both with their customers and with other dealers. The tier where dealers and customers trade is referred to as the customer market and the tier where dealers trade with other dealers is referred to as the interdealer market. A new data set including trades in both the interdealer and customer markets, as well as the identities of the buying and selling dealers allows for an investigation of how different dealers contribute in the price discovery process and of the type of information embedded in interdealer order flow. In sovereign bond markets, private information can be divided into types along two dimensions. The first dimension is whether information is fundamental or non-fundamental. Fundamental private information is related to macroeconomic factors, for example heterogeneous interpretations of macroeconomic indicators like consumer and producer surveys on future inflation, employment, and GDP growth. Non-fundamental private information is related to other factors like changes in liquidity conditions, investor risk preferences, auction volumes or hedging demands. Fundamental private information can influence expected future short rates and risk premia through the macroeconomic outlook, while non-fundamental private information can, for example, influence liquidity risk premia. The second dimension of private information is the source. Two sources of information are considered: dealer skill and customer trades.2 This paper defines dealer skill as a dealer's ability to acquire and interpret relevant information including public news and the order flow of other dealers. If the information in interdealer order flow reflects the information in customer order flow only, it indicates that the dealer does not add any information by trading. In this case, dealers are passive intermediaries of customer orders and customer trades are considered to be the source of information in interdealer order flow. If interdealer order flow is more informative than customer order flow, it suggests that dealers possess more information than their customers. Dealers can obtain extra information by using effort and skill in collecting and interpreting other relevant information. In this case, dealer skill is considered to be the source of information. Anand and Subrahmanyam (2008) find that dealers contribute more to price discovery than their customers and conclude that dealers are better informed than other market participants. In equity markets, insider information related to firm earnings or firm mergers may be a third source of information, but this type of information is unlikely to be of importance in sovereign bond markets. This study has two major contributions. The first is to compare the informational content of interdealer order flow to that of customer order flow to investigate in which tier of the market private information originates. Previous studies employ either interdealer order flow or customer order flow. Evans and Lyons (2002), Brandt and Kavajecz (2004), and Anand and Subrahmanyam (2008) use interdealer order flow from the bond, equity, and currency markets, respectively. Evans and Lyons (2005) use customer order flow from the currency market while Menkveld, Sarkar, and van der Wel (2012) use customer order flow from the bond futures market. This paper employs a unique data set including all trades in both tiers of the market, which makes it possible to investigate the price impact of interdealer order flow relative to customer order flow. The data set also identifies different types of trades. A type of particular interest in this study is delayed publication trades. These are trades chosen by dealers to be temporarily hidden from the other dealers. They become visible in the trading system only after a period of delay. Dealers are likely to choose this alternative if the trades contain private information as they can benefit from private information before it becomes available to their competitors. Delayed publication customer trades are therefore used as a proxy for informed customer trades. By using this proxy, I attempt to separate the effects of informed customer order flow from the effects of total customer order flow. The second contribution is to explore whether dealers are heterogeneous and play different roles in the price discovery process. Are dealers differently informed and do they have different sources of information? Are some dealers pure intermediaries while other dealers possess skill in collecting and interpreting relevant information that they subsequently trade on? This type of individual dealer analysis has not been undertaken so far due to a lack of data. Peiers (1997) and Sapp (2002) investigate information asymmetries between dealers in the foreign exchange market by using indicative quote data from currency dealers. They find that some dealers are price leaders and conclude that the results are consistent with news being incorporated into prices through trading among heterogeneously informed agents. However, as they use quote data instead of using transactions data directly they cannot document the trading behavior of the dealers who appear to have an informational advantage. The present study adds to the literature by employing individual dealer order flow in both the interdealer market and the customer market to explore where the information held by the informed dealers comes from. By studying the characteristics of each dealer, including market share, relative activity in the interdealer market, and the correlation between customer order flow and interdealer order flow, I infer whether the main source of information of the different dealers is customer order flow or dealer skill in acquiring and interpreting other relevant information. The results show that interdealer order flow in the Norwegian government bond market contains information about bond yields. Aggregate interdealer order flow explains one-fourth of daily yield changes in 3-, 5-, and 10-year government bonds. This is in line with the findings of Brandt and Kavajecz (2004). Order flow is divided into short-, medium-, and long-term order flow according to the remaining time to maturity of the bonds included.3 All three order flow groups have a significant impact on yield changes of all maturities, but the strongest effect is on the same maturity yield change. Short-term order flow thus has the highest impact on 3-year yield changes, but also has a significant impact on 5- and 10-year yield changes. These results indicate that there is information in the order flow of short-term, medium-term, and long-term bonds and that this information to a large extent causes parallel shifts in this part of the yield curve. The results further document that customer trades are far less informative than interdealer trades. Aggregate customer order flow explains up to 1% of daily variation in yields. The differences in the explanatory power of interdealer order flow and customer order flow suggest that dealers are better informed than their customers. When using a proxy for informed customer order flow, the explanatory power increases somewhat. This indicates that dealers, at least partially, are able to identify their informed customers. By aggregating information from their own customer trades, the trades of other dealers, and by processing public information related to bond markets, dealers have better access to relevant bond information than most customers.4 At the dealer level, the results show that dealers are heterogeneous and contribute differently to the price discovery process. The order flow of large dealers, measured by market share, has the largest price impact. The interdealer order flows of the two largest dealers have the highest price impact on 3- and 5-year bonds, whereas the order flows of the fourth largest dealer has the highest price impact on 10-year bonds. Also, order flows at different maturities have a different price impact depending on the dealer. Whereas the medium-term order flow has the highest price impact on all yield changes for one dealer, the long-term order flow has the highest price impact for other dealers. This suggests that dealers specialize in trading at different segments of the yield curve and therefore concentrate their informed interdealer trading to this segment. The observation that a dealer's interdealer order flow in one segment influences both short-, medium-, and long-term yields confirms the results at the aggregate level that much of the information contained in the order flow of government bonds is common for all the segments of the yield curve included in this study. The results at the dealer level also show that the correlation between customer order flow and interdealer order flow varies among dealers. Whereas customer order flow explains a substantial part of interdealer order flow for some dealers with high price impact, customer order flow appears to be unrelated to the interdealer order flow of other high-impact dealers, suggesting that some dealers possess information additional to what they learn from their customer trades. This is in line with the findings of Manaster and Mann (1996), who study the market for commodity futures, Anand and Subrahmanyam (2008), who study the equity market, and Osler, Mende, and Menkhoff (2011), who study the foreign exchange market. Additional information can be acquired through dealer skill. Examples of this type of information are the interpretation of early macroeconomic indicators or private information acquired through trading in the interdealer market. The findings in this study indicate that dealers play an important and independent role in the price formation process in the Norwegian government bond market as both customer trades and dealer skill appear to be important sources of information. The rest of the paper is organized as follows. Section 2 discusses related literature. Section 3 describes the market settings and data. Section 4 discusses possible dealer strategies. Section 5 presents the econometric framework and the results. Section 6 concludes.

This study explores the price discovery process in government bond markets by employing a new data set with complete trading records and dealer identities from the Norwegian government bond market. The main contributions are to compare the information content of interdealer order flow to that of customer order flow in a two-tier market and to explore whether price impacts vary across dealers. Previous studies on the price discovery process use more limited data sets. Osler, Mende, and Menkhoff (2011) use customer and interdealer trades from one dealer only, Brandt and Kavajecz (2004) use aggregate interdealer order flow only, while Manaster and Mann (1996), Anand and Subrahmanyam (2008), and Menkveld, Sarkar, and van der Wel (2012) employ data from one-tier markets. The results reveal that while aggregate interdealer order flow is highly informative, aggregate customer order flow is not. A proxy for informed customer trades, delayed publication trades, appear to contain some information indicating that at least some dealers are able to identify informed customers. The results further show that the interdealer flows of some dealers have a higher price impact than the interdealer order flow of others. These results are consistent with Peiers (1997) and Sapp (2002), who find that there are information asymmetries in the foreign exchange market and that some dealers are better informed than others. Dealers also appear to have different sources of information. While there is a strong link between the customer order flow and the interdealer order flow of some dealers, there is no such link for other dealers. This suggests that both customer trades and dealer skill in acquiring and interpreting other relevant information are sources of information in interdealer order flow. The results thus indicate that dealers in two-tier government bond markets are not mere intermediaries of customer trades, but play an independent role in the price discovery process. This is consistent with Osler, Mende, and Menkhoff (2011), who conclude that price discovery in the currency market mainly takes place in the interdealer market.