Jmcb Internet Appendix

Online Appendix (jmcb-internet-appendix)

When to trigger

• The main text is over the recommended 40 pages and material must move out without losing the argument
• Derivations, extra robustness, or data-construction detail are crowding the core narrative
• A referee will want to verify a step (a proof, a series construction) that does not belong in the body
• The replication/data question is unresolved and restricted bank/central-bank data complicate it
• The appendix has grown into a second paper with no map back to the main claims

The JMCB appendix logic

JMCB explicitly excludes the online appendix from the recommended 40-page limit (检索于 2026-06；以官网为准), which makes the appendix the pressure valve for a tight main text — but the journal's replication heritage means the appendix is also where verifiability lives. The discipline is twofold: the main text must remain self-contained (a reader who never opens the appendix still believes the result), and the appendix must map cleanly back (every appendix table answers a question the main text raised). Material that is load-bearing for the headline belongs in the body; material that supports or verifies belongs in the appendix.

What goes where

Replication and data path

• State the data/code plan early. Given JMCB's archive history, referees expect a clear answer on what will be deposited.
• Restricted data: if bank/central-bank/supervisory data cannot be shared, request a data exemption at submission (检索于 2026-06；以官网为准) and document the access path (RDC, central-bank data room, register) so a replicator could in principle obtain it.
• What the package can contain: code, constructed-variable definitions, crosswalks, and any non-restricted inputs; state explicitly what is withheld and why.
• Keep a table-by-table map from each appendix exhibit to the main-text claim it supports.

Structuring the appendix so referees use it

A well-organized appendix earns goodwill; a sprawling one frustrates the referee who has to mine it.

• Mirror the main-text order. Appendix Section B.3 should support main-text Section 3; a referee checking a claim should find its backup by section number, not by searching.
• Self-label every exhibit. Each appendix table/figure repeats the shock, units, sample, and specification — a referee opens the appendix out of order and must not have to flip back to decode it.
• Lead each appendix section with one sentence stating which main-text claim it supports and what threat it addresses.
• Keep proofs complete but signposted — state the proposition in the body, give the full proof in the appendix, and cross-reference both ways.

The replication question is a JMCB litmus test

Because JMCB's own archive history is part of the field's collective memory about reproducibility, treat the data/code plan as a first-class deliverable, not an afterthought. Decide, before submission, exactly which inputs are public (Call Reports, Y-9C, FRED/ALFRED vintages) and which are restricted, and write the access path for the restricted ones (RDC, central-bank data room, credit register, with the application route). If a step in the construction is non-obvious — a splice across a regime break, an M&A crosswalk, a winsorization rule — it belongs in the appendix in enough detail to rebuild. A referee who can see exactly how to reproduce the public part and exactly how to request the restricted part is far more likely to trust the result.