Benchmarks and software standards: A case study of GARCH procedures

McCullough, B. D.; Renfro, Charles G.

doi:10.3233/jem-1999-0160

Cited by 72 publications

(34 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To continue with the GARCH theme, McCullough and Renfro (1999) solved the same problem using seven packages and got seven results -see Tables 2.1 and 2.2. The t statistics for the coefficients were even more disparate than the coefficients.…”

Section: Inaccurate Garch Resultsmentioning

confidence: 99%

“…The original GARCH article (Bollerslev, 1986) neglected to explain some important details of the problem and the software developers simply had to guess at what Bollerslev did. McCullough and Renfro (1999) presented a benchmark based on the work of Fiorentini et al (1996). Fiorentini et al (FCP) provided the code for their implementation of the GARCH model, and FCP took especial pains to ensure the accuracy of their work.…”

Section: Inaccurate Garch Resultsmentioning

confidence: 99%

“…On what basis did McCullough and Renfro (1999) decide that the FCP code was of 'benchmark quality'? First, Calzolari and Panattoni (1988) had already proven that they could write benchmark-quality code when they provided a benchmark for FIML estimation.…”

Section: Inaccurate Garch Resultsmentioning

confidence: 99%

See 2 more Smart Citations

The Accuracy of Econometric Software

McCullough

2009

Handbook of Computational Econometrics

View full text Add to dashboard Cite

Section: Inaccurate Garch Resultsmentioning

confidence: 99%

Section: Inaccurate Garch Resultsmentioning

confidence: 99%

See 1 more Smart Citation

The Accuracy of Econometric Software

McCullough

2009

Handbook of Computational Econometrics

View full text Add to dashboard Cite

“…The JAE archive has the FCP FORTRAN program. McCullough and Renfro (1999) used these data and code to produce the FCP GARCH benchmark and applied it to seven packages, "Y1" through "Y7." The model is y t = µ + ∈ τ , where ∈ τ   Ψ t−1 ~ N(0, h t ) and h t = α 0 + α 1 ∈ τ−1 2 + β 1 h t−1 .…”

Section: The Need For Specific Benchmarksmentioning

confidence: 99%

The Numerical Reliability of Econometric Software

McCullough

Vinod

1999

Journal of Economic Literature

134

View full text Add to dashboard Cite

Numerous examples show that some econometric software packages contain serious flaws, and that users cannot safely assume that their software is accurate. A brief survey of the fundamentals of computer arithmetic discusses the sources of numerical error and emphasizes that computer arithmetic is not at all like pencil-and-paper arithmetic. Both users and developers of econometrics software should first pay attention to accuracy, and only later consider user-friendliness. Details are provided for assessing the accuracy of basic estimation routines, statistical distributions, and random number generators. More accuracy benchmarks are needed, especially for specialized econometric procedures.

show abstract

“…results from a large macroeconometric model because of difficulties in transferring programs and data across computer systems. McCullough and Renfro (1999), in a survey of GARCH estimation procedure implemented in various packages, found that often important information that affects computed results, such as parameter initialization, was not available. Buckheit and Donoho (1995) pointed out that in the field of computational experiments researchers often cannot reproduce their own work, even only a few months after its completion, that research students have difficulties in presenting their problems to their academic advisers, and that researchers cannot reproduce computational results of other researchers and other published work.…”

mentioning

confidence: 99%

Reproducible research in computational economics: guidelines, integrated approaches, and open source software

Baiocchi

2007

Comput Econ

View full text Add to dashboard Cite

Traditionally, computer and software applications have been used by economists to off-load otherwise complex or tedious tasks onto technology, freeing up time and intellect to address other, intellectually more rewarding, aspects of research. On the negative side, this increasing dependence on computers has resulted in research that has become increasingly difficult to replicate. In this paper, we propose some basic standards to improve the production and reporting of computational results in economics for the purpose of accuracy and reproducibility. In particular, we make recommendations on four aspects of the process: computational practice, published reporting, supporting documentation, and visualization. Also, we reflect on current developments in the practice of computing and visualization, such as integrated dynamic electronic documents, distributed computing systems, open source software, and their potential usefulness in making computational and empirical research in economics more easily reproducible. Copyright Springer Science+Business Media, LLC 2007Economic methodology, Econometric software, Other computer software, B4, C87, C88,

show abstract

Benchmarks and software standards: A case study of GARCH procedures

Cited by 72 publications

References 21 publications

The Accuracy of Econometric Software

The Accuracy of Econometric Software

The Numerical Reliability of Econometric Software

Reproducible research in computational economics: guidelines, integrated approaches, and open source software

Contact Info

Product

Resources

About