Julia_CT — Julia translation of Cameron & Trivedi's Microeconometrics Using Stata

Julia translations of the Stata code from Microeconometrics Using Stata (Cameron & Trivedi) as Quarto notebooks.

Files

• Cameron & Trivedi_stata.qmd — original Stata code (1st edition) as a Quarto notebook using stata_setup.
• Cameron & Trivedi_stata_vJulia.qmd — Julia translation of the Stata notebook above. Covers all 18 chapters + appendices.
• ch01_Stata_basics.qmd … ch18_Nonlinear_panel-data_models.qmd — per-chapter standalone notebooks generated from the master Julia qmd. Each is self-contained and renders independently (see Per-chapter notebooks below).
• split_chapters.jl — script that produces the per-chapter qmds from the master Julia qmd by extracting only the chunk-1 helpers each chapter transitively uses.
• annotate_with_stata.jl — script that prepends the corresponding Stata source from Cameron & Trivedi_stata.qmd to each Julia chunk in a chapter qmd, as a #= … =# reference comment.
• Cameron & Trivedi_stata_2ed_V2.qmd — 2nd-edition Stata notebook (work in progress).
• Cameron & Trivedi_stata_2ed_V2_testJulia.qmd — Julia test translation for the 2nd edition.
• Data folders (musr/, mus2/, mma10252005/, nhh2017cameron/) — sample datasets from the book's companion materials.
• auto.dta, census.dta, etc. — Stata built-in / sample data.

Per-chapter notebooks

In addition to the single master notebook (Cameron & Trivedi_stata_vJulia.qmd, ~38 k lines), the repo ships eighteen standalone chapter files — one per textbook chapter:

ch01_Stata_basics.qmd
ch02_Data_management_and_graphics.qmd
ch03_Linear_regression_basics.qmd
…
ch18_Nonlinear_panel-data_models.qmd

Each chapter file is self-contained and renders on its own. Its structure is:

1. YAML preamble (same as the master notebook).

2. Imports + compat-shims chunk — every using statement and the PrettyTables / Optim v3 compatibility shims.

3. Helpers chunk — only those chunk-1 helpers that this chapter actually references (transitively, via split_chapters.jl's dependency closure).

4. Chapter content — the original example chunks from the master notebook. Each Julia chunk is prepended with the corresponding Stata source from Cameron & Trivedi_stata.qmd as a #= … =# reference block, e.g.

   ```{julia}
   #=
   sysuse auto.dta
   summarize
   =#
   auto = DataFrame(ReadStatTables.readstat("auto.dta"))
   summarize_stata(auto)
   ```

This makes each chapter file useful as both a runnable Quarto notebook and a side-by-side Stata-↔-Julia reference.

Regenerating the chapter files

The per-chapter qmds are derived; the master file is the source of truth. To regenerate after editing Cameron & Trivedi_stata_vJulia.qmd:

julia --project=@. split_chapters.jl                   # rebuild ch01…ch18
for f in ch??_*.qmd; do                                # re-attach Stata refs
  julia --project=@. annotate_with_stata.jl "$f"
done

(Each annotate_with_stata.jl run also writes a *.qmd.bak the first time it touches a file, so the un-annotated version stays recoverable.)

Running

Stata notebook

Requires Stata 19 (or compatible) and stata_setup via Python:

pip install stata-setup

Render with:

quarto render "Cameron & Trivedi_stata.qmd"

Julia notebook

Requires Julia 1.10+ with these packages:

using Pkg
Pkg.add([
    "CairoMakie", "CategoricalArrays", "CSV", "Chain", "DataFrames",
    "Downloads", "FixedEffectModels", "GLM", "HypothesisTests",
    "PrettyTables", "ReadStatTables", "RegressionTables", "Statistics",
    "StatsBase", "CategoricalArrays", "LinearAlgebra", "KernelDensity",
    "Loess", "RDatasets", "Random", "Printf", "Optim", "Distributions",
    "SpecialFunctions", "MixedModels", "Combinatorics"
])

Render the master notebook with:

quarto render "Cameron & Trivedi_stata_vJulia.qmd"

Or render a single chapter:

quarto render ch04_Simulation.qmd

Scope

The Julia translation covers all 18 chapters of Microeconometrics Using Stata plus Appendices A and B.

Chapters

1. Stata basics — interactive use, command syntax, do-files, scalars/matrices, macros, looping, user-written commands.
2. Data management and graphics — data types, input/output, manipulation, reshaping, graphical display.
3. Linear regression basics — OLS in levels and logs, specification analysis, prediction, sampling weights, native Julia matrix-algebra OLS.
4. Simulation — pseudorandom-number generators, sample-mean distribution, computing integrals, simulation for regression.
5. GLS regression — GLS/FGLS, heteroskedastic models, systems of linear regressions (SUR), survey data (weights/clusters/strata).
6. Linear instrumental-variables regression — 2SLS, weak-instrument diagnostics, LIML/JIVE-style robust inference, 3SLS.
7. Quantile regression — QR for medical-expenditure data, generated heteroskedastic data, and count data (Machado–Santos Silva).
8. Linear panel-data models: Basics — pooled, within (FE), between, RE, first-difference, long panels, panel-data management.
9. Linear panel-data models: Extensions — panel IV, Hausman–Taylor (analytic/bootstrap/jackknife SEs), Arellano–Bond (difference GMM), mixed linear models, clustered data.
10. Nonlinear regression methods — nonlinear least squares and related estimators.
11. Nonlinear optimization methods — Optim.jl replacements for Stata's ml / Mata routines (Newton, BHHH, BFGS).
12. Testing methods — Wald, LR, LM/score, conditional moment, and non-nested tests.
13. Bootstrap methods — pairs/residual bootstrap, percentile-t, jackknife, bootstrap inference.
14. Binary outcome models — logit, probit, marginal effects, goodness-of-fit.
15. Multinomial models — multinomial / conditional / nested logit, multinomial probit via GHK simulator, mixed logit via simulated MLE (Halton draws).
16. Tobit and selection models — censored Tobit, Heckman two-step, sample-selection models.
17. Count-data models — Poisson, NegBin, hurdle, zero-inflated, finite mixtures.
18. Nonlinear panel models — FE logit, RE Tobit, FE/RE Poisson, panel NegBin.

Appendices

• A. Programming in Julia (replaces "Programming in Stata") — functions, control flow, modules.
• B. Native Julia matrix algebra (replaces Mata) — linear algebra patterns used throughout the book.

Testing status

The master notebook has been executed end-to-end and verified for all chapters.

All packages and helper functions used anywhere in the master notebook are loaded and defined in two upfront sections, so any chapter can be run independently after these are evaluated:

• Julia setup — top-of-notebook cell that calls using on every required package (CairoMakie, CategoricalArrays, CSV, Chain, DataFrames, Downloads, FixedEffectModels, GLM, HypothesisTests, PrettyTables, ReadStatTables, RegressionTables, Statistics, StatsBase, LinearAlgebra, KernelDensity, Loess, RDatasets, Random, Printf, Optim, Distributions, SpecialFunctions, MixedModels, Combinatorics).
• Chapter 1 — Stata basics — defines all reusable helper functions used by later chapters (table formatting, summary statistics, regression-output helpers, panel utilities, bootstrap/jackknife wrappers, etc.).

If you want to run only a specific chapter from the master notebook, evaluate the Julia setup cell and the helpers in Chapter 1 first. Alternatively, use the per-chapter standalone notebooks — each already includes the setup, only the helpers it needs, and the chapter's content in one self-contained file. Per-chapter render status is spot-checked (e.g. ch04_Simulation.qmd and ch17_Count-data_models.qmd have been rendered end-to-end); a full sweep of all 18 is in progress.

Notes

• PDFs of the textbooks are not included (copyright); please obtain them from the publisher.
• Rendered HTML outputs are not included; regenerate with quarto render.
• Bootstrap standard errors may differ slightly from Stata due to differing RNGs (Julia's MersenneTwister vs Stata's internal RNG). Point estimates match exactly.