Appendix f — Data Card — Mergent FISD (Fixed Income Securities Database)

Provider & what it is. Mergent FISD, distributed through WRDS, is the reference database of corporate-bond issue characteristics: coupon, maturity, offering amount and date, seniority, security/collateral, callability and other embedded options, covenants, and the agencies’ ratings history. If TRACE is the trade tape — what happened to a bond’s price — FISD is the bond’s biography — what the bond is. The reveal-the-trick point is that you almost never use FISD alone: it is the lookup table you join to a TRACE trade so that “this CUSIP traded at 101.25” becomes “a 10-year, BBB, callable, senior unsecured note traded at 101.25.” Most of the variables that explain a credit spread live in FISD, not in the price.

Coverage

Span. U.S. corporate and agency debt issues, with deep historical coverage of issues outstanding from roughly the 1990s forward and many earlier issues; rating histories from the agencies over the issue’s life. [CHECK] exact earliest reliable coverage and the universe of agency/MTN inclusion.
Frequency. Issue-level reference data (mostly static per issue), plus event tables that are time-stamped: rating changes, calls, and amendments. Not a price series — there is no daily frequency.
Universe. Corporate bonds, medium-term notes, and agency debt issued or registered for U.S. markets; both currently outstanding and matured/retired issues.

Key identifiers

CUSIP (9-character) — the issue-level join key to TRACE and to holdings data. As elsewhere, CUSIP is a licensed identifier; use it to merge, do not redistribute a master.
ISSUE_ID — FISD’s own internal issue identifier (stable within FISD; use it for joins within the FISD tables, e.g., issue → rating history).
ISSUER_ID / issuer CUSIP (6-char) — the FISD issuer key; one issuer has many issues. Bridge issuer to Compustat GVKEY via name/CUSIP matching — a Ch 7.4 crosswalk, and a messy one.

Access path

Same WRDS/wrds pattern; credentials from .pgpass or env (CONVENTIONS §5). FISD is split across an issue table and several event tables (ratings, etc.), joined on ISSUE_ID.

import os, wrds, pandas as pd, pathlib

db = wrds.Connection(wrds_username=os.environ["WRDS_USERNAME"])   # secret from env

raw = pathlib.Path("data/raw/fisd_issues.parquet")               # git-ignored
if raw.exists():
    issues = pd.read_parquet(raw)
else:
    issues = db.raw_sql("""
        SELECT issue_id, complete_cusip, issuer_cusip, offering_date,
               offering_amt, coupon, maturity, security_level,
               bond_type, callable, convertible
        FROM   fisd.fisd_mergedissue
        WHERE  bond_type IN ('CDEB','CMTN')        -- corp debentures / MTNs
    """, date_cols=["offering_date", "maturity"])
    issues.to_parquet(raw)                                       # GMU infra only

    # Rating history is a SEPARATE event table, joined on issue_id:
    ratings = db.raw_sql("""
        SELECT issue_id, rating_date, rating, rating_type
        FROM   fisd.fisd_ratings
    """, date_cols=["rating_date"])
db.close()

[CHECK] the exact WRDS table names (fisd.fisd_mergedissue, fisd.fisd_ratings, the issuer table) — naming has varied across vintages; confirm with db.list_tables(library="fisd") and db.describe_table(...).

License & the “stays on GMU infrastructure” note (CONVENTIONS §5)

Mergent FISD is licensed (via WRDS), accessed read-only. Query only on WRDS Cloud or GMU’s Hopper, cache to a git-ignored data/raw/, and never commit, email off campus, post publicly, or paste into a commercial web service. CUSIP is separately licensed — no CUSIP master leaves the building. As with every WRDS source, the repository ships the query and the pull log; a GMU-seated collaborator re-runs the recipe to regenerate the extract.

Gotchas

It is reference data, not prices. FISD tells you what a bond is; pair it with TRACE for what it did. The whole point is the join.
Issue-level vs. issuer-level. One issuer (issuer_id) has many issues; aggregating to the issuer requires deciding how to weight (by offering amount? equally?). State the choice.
bond_type filtering. FISD mixes corporate debentures, MTNs, convertibles, asset-backed, and more. A “corporate bond” sample needs an explicit bond_type filter; including convertibles or ABS by accident contaminates a credit-spread study.
Ratings: which agency, which scale, and timing. The ratings table has multiple agencies and rating types; you must map letter grades to a numeric scale yourself and respect the rating_date so you use the rating in force at your event, not the latest one (a look-ahead trap).
Embedded options change the analysis. A callable or convertible bond’s spread is not comparable to a straight bond’s; the callable/convertible flags are not optional context — they change which bonds belong in your sample.
Maturity/retirement. A matured or called bond stops trading; align FISD’s maturity/call events with the TRACE window.

“First 10 rows” schema sketch (illustrative — not real FISD data)

issue_id (int)	complete_cusip (str)	issuer_cusip (str)	offering_date (date)	offering_amt ($000s)	coupon (float)	maturity (date)	security_level (str)	callable (str)
480211	037833AK6	037833	2013-05-03	2000000	2.400	2023-05-03	SEN	N
480212	037833AL4	037833	2013-05-03	3000000	3.850	2043-05-03	SEN	Y
512744	594918AC8	594918	2012-11-07	750000	2.125	2022-11-07	SEN	N
601233	17275RAH5	17275R	2015-03-03	1500000	3.450	2025-03-03	SUB	Y
…	…	…	…	…	…	…	…	…

(Values invented to show shape and types: rows 1–2 are two issues from the same issuer with different maturities and call features — why you cannot collapse an issuer to “one bond.”)

Which chapter/lab/capstone uses it

Week 7, Ch 7.2 — part of the WRDS fixed-income toolkit; pairs with the TRACE card.
The standard companion to TRACE for any fixed-income capstone (credit-spread cross-section, bond liquidity, rating-change event studies): FISD supplies the issue characteristics and rating history that the controls and sample filters require — an advanced extension beyond the equity-focused anomaly thread.