Public Micro Data

Key Concepts:

• N-Port data: Publicly available microdata on financial positions of investment management companies in the US.
• Accession file: Contains the identity of a portfolio report.
• Holding ID file: Contains characteristics of the securities held in the portfolio.
• Flow of Funds: National statistics of the US that look at various sectors of the economy and other related to each other and so they include positions.
• Q-SIP: Identifiers of securities with nine digits.
• Home Bias: The tendency of investors to hold a disproportionate amount of securities issued by their domestic firms.
• Currency Bias: The tendency of investors to hold a disproportionate amount of securities denominated in their domestic currency.

1. Introduction and Overview

• The session focuses on new, freely available public microdata, particularly relevant for PhD students with limited budgets but ample time.
• The availability of this data facilitates faster project initiation without lengthy contract negotiations and funding acquisition.
• The primary focus is on the N-Port data, a US government project requiring investment management companies to disclose their holdings quarterly.
• Bruno Kavani and Jesse are credited as the key figures in compiling and organizing this data.

2. N-Port Data: Details and Structure

• N-Port data is quarterly and covers only mutual funds, excluding hedge funds and other investor types.
• It provides detailed portfolio reports with holdings, currency information, and full identifiers (CUSIPs, ISINs).
• The data is organized into two key files: the accession file (portfolio report identity) and the holding ID file (security characteristics).
• The holding ID file includes crucial security characteristics such as type (corporate bond, equity, derivative), currency, maturity, and coupon.
• The speaker emphasizes the significant effort typically required to gather and clean this information, highlighting the value of the prepackaged data.

3. GitHub Repository and Public Good

• The GCAP lab is creating a GitHub repository with code for downloading and cleaning the N-Port data.
• The goal is to provide a ready-to-use dataset for analysis in software like Stata.
• The project is open-source, encouraging community engagement and feedback for identifying and fixing errors.

4. Data Example and Key Fields

• An example is provided showing holdings of the Vanguard Total Stock Market Index Fund (Microsoft, Apple) and the Vanguard Total International Stock Index Fund (Alibaba).
• Key fields include fund identifier, security identifier (ISIN, CUSIP), currency, and currency value.
• The example illustrates how the data reveals the country of issuance (e.g., Cayman Islands for Alibaba) and listing exchange (e.g., New York Stock Exchange).

5. Data Validation and Benchmarking

• A crucial step is benchmarking the microdata against aggregate statistics to assess its quality and coverage.
• The speaker compares N-Port data to the Flow of Funds data, showing a good match for total assets owned by mutual funds.
• Discrepancies are noted at finer levels of disaggregation (e.g., treasuries), emphasizing the need for caution.
• The US lacks a comprehensive administrative dataset for domestic holdings, making the N-Port data particularly valuable.
• Cross-border holdings data in the US is excellent due to strict reporting requirements.

6. Comparison with Commercial Data

• N-Port data is compared to commercial data (e.g., Morningstar) at the security level (ISIN).
• The agreement is remarkably close, especially for large positions.
• Differences exist, particularly for smaller positions, where misreporting is more common.

7. Data Limitations and Considerations

• The data covers only mutual funds, not the entire universe of investors.
• Non-bank financial intermediation has grown significantly, making mutual funds a dominant sector in terms of positions.
• Insurance companies are large holders of certain assets (e.g., corporate bonds), requiring consideration depending on the market being analyzed.
• As data is split into finer categories, definition discrepancies become more important.

8. Currency Analysis and Home Bias

• The speaker discusses analyzing foreign debt holdings by currency, comparing N-Port data to TIC disclosures.
• US funds hold a significant amount of dollars abroad (75% of holdings).
• The analysis extends to comparing mutual fund holdings to full aggregates, revealing differences due to the asset-liability matching of insurance companies.
• The concept of home bias is introduced, where investors hold more securities issued by their domestic firms.

9. Replicating Existing Research with N-Port Data

• The speaker demonstrates how to replicate existing research using the N-Port data.
• The example focuses on a paper examining the fraction of a security held by a US investor based on currency.
• The original paper used total holdings of all funds as the denominator, while the N-Port analysis uses total outstanding securities due to data limitations.
• The results show a large currency effect, consistent with the original findings.
• The home bias effect is also replicated, demonstrating the robustness of the finding across different datasets.

10. Q&A Highlights

• Data Provider Comparison: There are differences between N-Port and commercial data providers like Morningstar, but convergence is expected over time.
• Firm Identifiers: The data includes LEIs, CUSIPs, and ISINs, allowing for merging with other datasets and crosswalks.
• Data Frequency: The data is quarterly since 2019. Commercial data may have monthly data, but quality varies.
• Coverage: N-Port covers a large portion of the mutual fund universe, but not all investors.
• Derivatives: Derivatives coverage is limited and complex, with challenges in transforming notional amounts into equivalent exposure numbers.
• Short-Maturity Securities: The data provides a snapshot of holdings at the reporting date, so short-maturity securities are included if held at that time.

11. Conclusion

• The N-Port data represents a significant advancement in the availability of free, usable microdata for financial research.
• It can save researchers significant time and effort in data acquisition and cleaning.
• The speaker encourages community engagement to improve the data and cleaning codes.
• The data is particularly valuable for cross-sectional analysis and replicating existing research.