Scope and Limitations
This skill is a code pattern library, not a methodological advisor.
| Can Do |
Cannot Do |
| Show how published papers implemented methods |
Explain when to use one method over another |
| Provide tested STATA syntax |
Advise on identification strategy |
| Indicate which robustness tests accompany analyses |
Discuss research design trade-offs |
| Cite source papers for code patterns |
Recommend optimal research design |
When users ask methodology questions (e.g., "Should I use entropy balancing or PSM?", "How do I address endogeneity?", "Is my identification strategy valid?"):
- Acknowledge the limitation: "This skill provides code patterns from published papers, not research design guidance."
- Show how different papers approached similar problems (code examples)
- Suggest consulting methodology references: Breuer & deHaan (2024) for fixed effects, Angrist & Pischke for causal inference, or the user's methodologist/advisor
- Offer to show multiple implementations so the user can see variation in approaches
Workflow
Use references/REFERENCES.md as the primary index, then read targeted .do files.
Stage 1: Index Search
Search references/REFERENCES.md to identify relevant papers. The index contains structured metadata:
- Primary Method: STATA commands used (reghdfe, psmatch2, stcox, etc.)
- Identification Strategy: DiD, PSM, IV, RDD, Event Study, etc.
- Robustness/Special Features: Winsorization levels, clustering specs, placebo tests, etc.
Example queries on REFERENCES.md:
- "entropy balancing" → finds JAR_60_alv, JAR_60_bl, JAR_61_ds, JAR_62_5_llz, JAR_63_2_npstv
- "stacked DiD" → finds JAR_61_ds, JAR_62_5_aov, JAR_62_5_gibbons
- "Cox hazard" → finds JAR_59_ctv, JAR_62_2_xyz
Stage 2: Code Extraction
Read only the identified .do files to extract actual syntax. This reduces context usage and improves accuracy.
Stage 3: Adaptation and Citation
- Adapt patterns to the user's variable names and research context
- Cite source: "Based on [Authors] ([Year]), JAR Volume"
Fallback: Direct Grep Patterns
For very specific syntax queries (e.g., "how does absorb() handle singletons?"), grep .do files directly:
| Task |
Grep Pattern |
| Panel regressions |
reghdfe|xtreg|areg |
| Fixed effects |
absorb\(|i\.year|i\.firm |
| Clustering |
cluster\(|vce\(cluster |
| Matching/PSM |
psmatch2|teffects|cem|ebalance|pscore |
| IV regression |
xtivreg|ivregress|ivreg2 |
| DiD |
post.*treat|treat.*post|parallel.*trend |
| RDD |
rdrobust|rddensity |
| Event studies |
CAR|BHAR|abnormal.*return |
| Survival |
stcox|streg|stset |
| Fama-MacBeth |
fama.?macbeth|newey.*west |
| Bootstrap |
bootstrap|bsample |
| Quantile regression |
qreg|sqreg|bsqreg |
| Table output |
esttab|outreg2|eststo |
| Winsorization |
winsor|winsor2 |
Corpus Overview
126 STATA .do files from JAR Volumes 55-63 (2017-2025). See references/REFERENCES.md for complete catalog with paper titles and authors.
File Naming Convention
- V55-61:
JAR_{volume}_{shortcode}.do
- V62-63:
JAR_{volume}_{issue}_{shortcode}_{authors}.do
Volume Coverage
| Volume |
Year |
Papers |
| 55 |
2017 |
9 |
| 56 |
2018 |
12 |
| 57 |
2019 |
9 |
| 58 |
2020 |
13 |
| 59 |
2021 |
4 |
| 60 |
2022 |
22 |
| 61 |
2023 |
22 |
| 62 |
2024 |
25 |
| 63 |
2025 |
10 |
Standard Patterns
Clustering and Fixed Effects
* Firm and year FE with firm-clustered SEs (most common)
reghdfe depvar indepvar controls, absorb(firm year) cluster(firm)
* Industry-year FE
reghdfe depvar indepvar controls, absorb(ind_year) cluster(firm)
Output Conventions
eststo clear
eststo: reghdfe depvar indepvar controls, absorb(firm year) cluster(firm)
esttab using "table.tex", replace star(* 0.10 ** 0.05 *** 0.01) se
Winsorization
winsor2 varlist, cuts(1 99) replace
1---2name: stata-accounting-research3description: STATA code pattern library for empirical archival accounting research. Provides tested syntax from 126 peer-reviewed JAR (Journal of Accounting Research) replication files (2017-2025). Use when the user asks procedural questions like "How do I implement [method]?" or "Show me code for [technique]" — including: entropy balancing, propensity score matching (PSM), difference-in-differences (DiD), regression discontinuity (RDD), instrumental variables (IV), event studies (CAR/BHAR), survival analysis, Fama-MacBeth regressions, bootstrap, quantile regression, reghdfe/xtreg/areg, clustering standard errors, fixed effects, esttab/outreg2 table formatting, winsorization, leads/lags. Users can specify their variables (e.g., treatment, outcomes, controls) and receive adapted syntax. NOTE: This skill provides code patterns from published papers, not research design advice.4---56## Scope and Limitations78This skill is a **code pattern library**, not a methodological advisor.910| Can Do | Cannot Do |11|--------|-----------|12| Show *how* published papers implemented methods | Explain *when* to use one method over another |13| Provide tested STATA syntax | Advise on identification strategy |14| Indicate which robustness tests accompany analyses | Discuss research design trade-offs |15| Cite source papers for code patterns | Recommend optimal research design |1617**When users ask methodology questions** (e.g., "Should I use entropy balancing or PSM?", "How do I address endogeneity?", "Is my identification strategy valid?"):18191. Acknowledge the limitation: "This skill provides code patterns from published papers, not research design guidance."202. Show how different papers approached similar problems (code examples)213. Suggest consulting methodology references: Breuer & deHaan (2024) for fixed effects, Angrist & Pischke for causal inference, or the user's methodologist/advisor224. Offer to show multiple implementations so the user can see variation in approaches2324## Workflow2526Use `references/REFERENCES.md` as the primary index, then read targeted .do files.2728### Stage 1: Index Search2930Search `references/REFERENCES.md` to identify relevant papers. The index contains structured metadata:31- **Primary Method**: STATA commands used (reghdfe, psmatch2, stcox, etc.)32- **Identification Strategy**: DiD, PSM, IV, RDD, Event Study, etc.33- **Robustness/Special Features**: Winsorization levels, clustering specs, placebo tests, etc.3435Example queries on REFERENCES.md:36- "entropy balancing" → finds JAR_60_alv, JAR_60_bl, JAR_61_ds, JAR_62_5_llz, JAR_63_2_npstv37- "stacked DiD" → finds JAR_61_ds, JAR_62_5_aov, JAR_62_5_gibbons38- "Cox hazard" → finds JAR_59_ctv, JAR_62_2_xyz3940### Stage 2: Code Extraction4142Read only the identified .do files to extract actual syntax. This reduces context usage and improves accuracy.4344### Stage 3: Adaptation and Citation45461. Adapt patterns to the user's variable names and research context472. Cite source: "Based on [Authors] ([Year]), JAR [Volume]([Issue])"4849## Fallback: Direct Grep Patterns5051For very specific syntax queries (e.g., "how does absorb() handle singletons?"), grep .do files directly:5253| Task | Grep Pattern |54|------|--------------|55| Panel regressions | `reghdfe\|xtreg\|areg` |56| Fixed effects | `absorb\(\|i\.year\|i\.firm` |57| Clustering | `cluster\(\|vce\(cluster` |58| Matching/PSM | `psmatch2\|teffects\|cem\|ebalance\|pscore` |59| IV regression | `xtivreg\|ivregress\|ivreg2` |60| DiD | `post.*treat\|treat.*post\|parallel.*trend` |61| RDD | `rdrobust\|rddensity` |62| Event studies | `CAR\|BHAR\|abnormal.*return` |63| Survival | `stcox\|streg\|stset` |64| Fama-MacBeth | `fama.?macbeth\|newey.*west` |65| Bootstrap | `bootstrap\|bsample` |66| Quantile regression | `qreg\|sqreg\|bsqreg` |67| Table output | `esttab\|outreg2\|eststo` |68| Winsorization | `winsor\|winsor2` |6970## Corpus Overview7172126 STATA .do files from JAR Volumes 55-63 (2017-2025). See `references/REFERENCES.md` for complete catalog with paper titles and authors.7374### File Naming Convention7576- V55-61: `JAR_{volume}_{shortcode}.do`77- V62-63: `JAR_{volume}_{issue}_{shortcode}_{authors}.do`7879### Volume Coverage8081| Volume | Year | Papers |82|--------|------|--------|83| 55 | 2017 | 9 |84| 56 | 2018 | 12 |85| 57 | 2019 | 9 |86| 58 | 2020 | 13 |87| 59 | 2021 | 4 |88| 60 | 2022 | 22 |89| 61 | 2023 | 22 |90| 62 | 2024 | 25 |91| 63 | 2025 | 10 |9293## Standard Patterns9495### Clustering and Fixed Effects9697```stata98* Firm and year FE with firm-clustered SEs (most common)99reghdfe depvar indepvar controls, absorb(firm year) cluster(firm)100101* Industry-year FE102reghdfe depvar indepvar controls, absorb(ind_year) cluster(firm)103```104105### Output Conventions106107```stata108eststo clear109eststo: reghdfe depvar indepvar controls, absorb(firm year) cluster(firm)110esttab using "table.tex", replace star(* 0.10 ** 0.05 *** 0.01) se111```112113### Winsorization114115```stata116winsor2 varlist, cuts(1 99) replace117```
brycewang-stanford