NEWS.md

In badp: Bayesian Averaging for Dynamic Panels

badp 0.6.0

• Replaced the C++ (Rcpp/RcppArmadillo) SEM likelihood implementation with an R implementation differentiated via RTMB (automatic differentiation). Optimization now uses gradients obtained by automatic differentiation instead of finite differences, and standard errors are computed from the Hessian and per-entity score vectors of the same tape instead of finite-difference approximations. Optimized parameters and standard errors therefore differ from those of earlier versions; likelihood values at given parameters are unchanged. The difference is small for the parameter estimates but can be substantial for the robust standard errors, which depend on the derivatives twice over. The finite-difference hessian() function was removed. The Rcpp, RcppArmadillo, rootSolve and optimbase dependencies were dropped in favour of RTMB.
• init_value (in optim_model_space() and related functions) now also accepts a generator function of one argument n returning n starting values (e.g. function(n) runif(n, 0.1, 1)), enabling randomized multi-start experiments, or a single number used as the starting value for every parameter, as before (init_value = 0.5 is equivalent to function(n) rep(0.5, n)), so code written against 0.4.0 and 0.5.0 continues to work. Passing 0, a vector of length greater than one, or anything that is neither a function nor a single finite number now fails with an informative error, as does a generator that draws exactly 0 for an included parameter; zero is rejected either way because it is reserved to mark a parameter excluded from a given model.

• Per-model optimization is more robust to the harder cases automatic differentiation now makes tractable to explore:

  • BFGS is restarted from its own solution until the log-likelihood value stops improving by more than restart_tol (max_restarts and restart_tol arguments of optim_model_space()); with max_restarts = 0, optim()'s own convergence flag is used directly.
  • A starting point at which the likelihood is undefined, and an automatic-differentiation error encountered mid-optimization, no longer abort the procedure: the offending point is treated as a rejected step, letting the line search shrink it, or, if it was the initial point, redrawn from init_value (up to max_init_attempts times).
  • A model whose observed information matrix is not positive definite, or not finite, is re-optimized from a freshly drawn starting point (up to max_reoptimizations times) instead of being reported as a failure.
  • Per-model convergence diagnostics - converged flag, optim code, number of restarts, number of initial draws, and final gradient norm - are stored in the new convergence element of badp_model_space objects. Non-converged models trigger a warning in optim_model_space() and bma() and are reported by summary()/print(); they are deliberately not excluded from the analysis.

• The model space statistics gained a row holding the numerical rank of J = sum_i s_i s_i', the outer product of the entity-level scores from which the robust ("sandwich") standard errors are built. Because the per-entity scores share a component identical across entities, and because they sum to zero at the maximum, J is unchanged by centring them and depends only on the variation of the scores across entities. Parameters entering the log-likelihood solely through terms common to every entity contribute nothing, so J is rank deficient however many entities are observed. In the bundled model spaces the scores span between 8% and 29% of the parameter directions.

The robust standard deviations that bma() reports are unaffected in the sense that they remain well defined: J vanishes outside the spanned block, so the sandwich restricted to that block equals the profile sandwich obtained by profiling the remaining parameters out. What the construction discards is the score covariance involving those remaining directions, which is set to zero rather than estimated. The likelihood, the posterior means, the posterior inclusion probabilities and the Hessian-based standard errors PSD and PSDcon do not involve J.

summary() of a model space now reports the fraction of parameter directions spanned. bma() gained an eta argument taking the learning rate directly, which overrides weighting. eta = 1 reproduces "mb2012" and eta = 1/N reproduces "mb2016", so the sensitivity of any conclusion to the rate can be examined without re-estimation. The "curvature" weighting, which estimated the learning rate by the magnitude ("omnibus") adjustment for misspecified likelihoods, has been withdrawn before release. The adjustment assumes J estimates the variance of the score, and J is rank deficient for this likelihood, so the rate would be calibrated on the small part of the parameter space the entity-level scores span, with no way to assess what the remainder contributes. Use eta to set a rate explicitly instead. The ingredients, tr(H^-1 J), dim(theta) and rank(J), remain stored with every fitted model space, so the rate can still be computed and inspected directly. * bma() gained a weighting argument selecting the approximation to the marginal likelihood used to weight the models. Writing A_j = loglik_j - (k_j/2)*log(N*T), three of the four options are the same construction with different learning rates eta, log w_j = eta * A_j: * "mb2016" (default, eta = 1/N) is unchanged behaviour, the approximation computed by the implementation accompanying Moral-Benito (2016), and the option that reproduces the posterior inclusion probabilities and posterior moments published there; * "mb2012" (eta = 1) is the Schwarz criterion exactly as stated in equations (24)-(30) of Moral-Benito (2012); * "nt" (eta = 1/(N*T)) averages over entity-periods rather than entities, the scaling that would be internally consistent with the log(N*T) penalty.

The fourth option, "uip", instead alters the penalty, using exp(loglik - (k/2)*log(N)) with the entity as the unit of information, as implied by the unit information prior of Kass and Wasserman (1995) given that the likelihood factorises over entities.

Any eta != 1 gives a tempered (power) posterior over models, which lies outside the approximation that motivates the criterion. A rate held fixed as the sample grows only rescales the log weights, so the posterior still concentrates, more slowly for eta < 1. A rate that shrinks with the sample is different in kind: under eta = 1/N the log weights converge to constants and the posterior never concentrates. Conversely eta = 1 concentrates sharply and can place nearly all posterior mass on a single model. The choice can therefore change posterior inclusion probabilities materially, and users are encouraged to check the sensitivity of their conclusions. Switching between the options requires no re-estimation, as all are recovered from the same fitted model space. The selected weighting and the realised learning rate are recorded in the new weighting and eta elements (slots 18 and 19) of badp_bma objects. Existing slots 1-17 are unchanged, so numeric indexing of earlier elements continues to work. Model weights are now formed on the log scale and shifted before exponentiation, which prevents overflow when log Bayes factors are large. coef_hist() labels the y axis "Density" rather than "Frequency", matching what the histograms actually show. sem_sigma_matrix() builds variances with multiplication rather than ^2, avoiding a NaN second derivative that automatic differentiation would otherwise produce when a variance parameter is exactly zero. sem_C_matrix() validates that phi_1 is supplied with the same length as beta. RTMB (>= 1.6) is now required, for automatic-differentiation support in chol() and determinant(). The minimum required R version was raised from 3.5 to 4.4, to match the requirement of Matrix, on which RTMB depends through TMB. The previous declaration could not be satisfied in practice. Regenerated every bundled model space (small_model_space, full_model_space, model_space_nonnested, migration_model_space and migration_model_space_nonnested) and full_bma_results with the fixed automatic differentiation pipeline. model_space_nonnested had no generating script; data-raw/model_space_nonnested.R now provides one.

badp 0.5.0

• Breaking change: best_models() now takes a character prior argument in place of the integer criterion argument. Use prior = "binomial" (default) instead of criterion = 1, and prior = "beta" instead of criterion = 2. This brings the API in line with summary.badp_bma(), which already used prior = "binomial" | "beta".
• Breaking change: Renamed dil.Par parameter to omega for clarity and consistency with statistical literature.
• Added S3 classes and methods for JSS compliance:
  • bma() now returns an object of class badp_bma (previously unclassed list).
  • optim_model_space() now returns an object of class badp_model_space.
  • Implemented S3 methods for badp_bma objects:
    • print.badp_bma() - Clean, informative console output.
    • summary.badp_bma() - Detailed statistical summary with highlighted important variables. Enhanced to display BMA statistics for both binomial and binomial-beta priors simultaneously.
    • coef.badp_bma() - Extract coefficients with optional standard errors and PIPs.
    • plot.badp_bma() - Default visualization with dispatch to existing plot functions.
  • Implemented print.badp_model_space() for model space objects.
  • Fixed component names in bma() output: removed spaces, duplicates, and typos; all names are now valid R identifiers (e.g., uniform_table, random_table, reg_names, dilution, alphas).
  • Compatibility note: Numeric indexing (results[[3]]) and helper functions (best_models(), jointness(), etc.) are fully preserved. Named access is available via the new identifiers (e.g., results$reg_names), but code using the previous long component names must be updated.
  • Added comprehensive tests for S3 methods and for preserved numeric-indexing/helper-function compatibility (125 new tests).
• Improved documentation: Added @keywords internal to hide helper and implementation functions from user-facing help documentation.
• Replaced sem_likelihood example: use the bundled economic_growth dataset instead of small random data that could produce NA or invalid positive values on some platforms.
• Removed ggpubr dependency; plotting functions now use patchwork for plot arrangement.
• Added migration_data dataset with migration flows data from Afonso, Alves, & Beck (2025).
• Added migration_model_space and migration_model_space_nonnested example model space objects.
• Fixed feature_standardization function to handle tibble input correctly.
• Exported join_lagged_col function.
• Standardized internal variable naming to R-idiomatic conventions (e.g., n_ prefix for counts, df_free for degrees of freedom).
• Fixed spelling mistakes and grammar in documentation.
• Added a devbox-based reproducible development environment (devbox.json) and a justfile with shortcuts for common development tasks (just test, just check, just document, etc.).

badp 0.4.0

• Renamed package from bdsm to badp (Bayesian Averaging for Dynamic Panels).
• Removed the df argument from the bma function; data is no longer required at the BMA stage.
• Added posterior_dens function for plotting posterior densities of coefficients.
• Added weighted coefficient histograms in coef_hist via the weight parameter (based on posterior model probabilities).
• Exported extract_names function.
• Recomputed bundled datasets to be consistent with updated optim_model_space.

bdsm 0.3.0

• Reimplemented SEM likelihood computation in C++.

bdsm 0.2.2

• Modified the method for selecting beta coefficient rows in the bma function for improved robustness and compatibility.
• Updated tests to align with changes in the upcoming ggplot2 release (v4.0.0), ensuring compatibility and future-proofing the package.

bdsm 0.2.1

• Added a vignette explaining Bayesian model averaging for dynamic panels with weakly exogenous regressors

bdsm 0.2.0

• Added GitHub Actions Workflows:
  • .github/workflows/R-CMD-check-develop.yaml: A workflow for R CMD checks on the develop branch.
  • .github/workflows/R-CMD-check-main.yaml: A workflow for R CMD checks across multiple operating systems and R versions on the main branch.
• Updated .Rbuildignore:
  • Ignored the .github directory.
• Updated .gitignore:
  • Added rules to ignore R-specific temporary files, build outputs, and vignettes.
• Updated DESCRIPTION:
  • Added rmarkdown and pbapply to Suggested and Imports, respectively.
  • Updated the dependency on R to version >= 3.5.
• Updated NAMESPACE:
  • Adjusted function exports to follow naming conventions (e.g., SEM_ functions renamed to sem_).
• Re-factored R Functions:
  • Renamed SEM_ functions to sem_ in multiple files for consistency.
• Removed R/SEM_bma.R:
  • The file R/SEM_bma.R was deleted, indicating major re-factoring or deprecation of related functionality.
• Added progress bar for computationally intensive functions
• Changed naming convention and broadened the meaning of a model space. Now it is a list containing two named elements: parameters (params) of all considered models and statistics (stats) computed using these parameters. This is a much more comprehensible naming convention than the previous one, where only the parameters were considered as the model space. Along with that change, some re-factoring and modifications were introduced:
  • all functions relating to the model space are now stored in R/model_space.R
  • initialize_model_space was renamed to init_model_space_params
  • likelihoods_summary was renamed to compute_model_space_stats
  • optimal_model_space was renamed to optim_model_space_params
  • a wrapper function optim_model_space, which returns the entire model space (both parameters and statistics), was introduced
  • data objects released with the package were re-factored, recomputed, and renamed. Two example model spaces computed with the new optim_model_space function are provided: small_model_space and full_model_space.
• Simplified the framework for data preparation. A single function feature_standardization is provided, which allows flexible and simple options for data preparation. See the vignette and function manual for more details.

bdsm 0.1.0

• Initial CRAN submission.

badp documentation built on Aug. 20, 2026, 9:08 a.m.