X-12-ARIMA

X-12-ARIMA
Name	X-12-ARIMA
Developer	U.S. Census Bureau
Released	1990s
Latest release	Legacy (superseded by newer packages)
Operating system	Unix, Microsoft Windows, macOS
Programming language	Fortran (programming language), C (programming language)
Genre	Seasonal adjustment, Time series analysis
License	Public domain (historical)

X-12-ARIMA is a statistical seasonal adjustment and time series decomposition package developed by the U.S. Census Bureau for analyzing monthly and quarterly data. It extended earlier work associated with Tramo-SEATS research and builds on methodologies linked to Box–Jenkins methodology, George E. P. Box, Gwilym Jenkins, and instrumented practices used by agencies such as the Federal Reserve Board and the International Monetary Fund. Widely used in official statistics and applied research, it influenced later systems employed by institutions including the Organisation for Economic Co-operation and Development, the World Bank, and the Bureau of Labor Statistics.

History

X-12-ARIMA evolved from earlier seasonal adjustment systems like the Census X-11 family and predecessors designed by the U.S. Census Bureau and researchers associated with Ernest T. A. Koopmans-era approaches. Development incorporated advances from scholars such as Box–Jenkins methodology proponents including George E. P. Box and Gwilym Jenkins and drew on work by Gareth W. Jones and researchers linked to Tramo-SEATS initiatives. The package formalized procedures during the late 20th century when institutions like the Bureau of Economic Analysis and Eurostat sought reproducible seasonal adjustment for national accounts. X-12-ARIMA became a standard tool within agencies including the Bank of England, the Reserve Bank of Australia, and the Statistics Canada environment before later being superseded in many workflows by successors developed by the U.S. Census Bureau and open-source communities.

Methodology

X-12-ARIMA implements a model-based seasonal adjustment workflow influenced by Box–Jenkins methodology for ARIMA model identification and forecasting and the additive and multiplicative decomposition frameworks used by statistical offices such as Eurostat and OECD. Its methodology combines prefiltering with ARIMA regression modeling concepts advanced by researchers active at the U.S. Census Bureau and aligns with standards promoted by organizations like the United Nations Statistical Division and the International Labour Organization for time series treatment. Analysts commonly compare its outputs with results from procedures developed by J. M. Peña and Victor Gomez and algorithmic approaches from teams at Banco de España and central statistical agencies including Instituto Nacional de Estadística (Spain).

Features and Components

X-12-ARIMA contains modular components for outlier detection, trading day adjustment, and Easter effect modeling that echo procedural elements used by agencies such as the Bureau of Labor Statistics and the Office for National Statistics. It includes automated ARIMA model selection inspired by Box–Jenkins methodology and diagnostic testing comparable to tests used in research by Clive Granger and Robert F. Engle. The software produces diagnostics and visual outputs used in policy settings within institutions like the European Central Bank and the International Monetary Fund. Key components mirror statistical conventions present in publications from National Bureau of Economic Research authors and align with series treatment standards from the Organisation for Economic Co-operation and Development.

Applications

X-12-ARIMA was applied in official-statistics production at the U.S. Census Bureau, Bureau of Labor Statistics, and Statistics Canada for adjusting indicators such as industrial production, retail sales, and employment series reported by entities like the U.S. Bureau of Economic Analysis and the Office for National Statistics (UK). Researchers at universities including Harvard University, Massachusetts Institute of Technology, London School of Economics, and University of Chicago employed it in empirical macroeconomic studies and forecasting exercises relevant to central banks such as the Federal Reserve Board, the European Central Bank, and the Bank of Japan. International organizations including the World Bank, the International Monetary Fund, and Eurostat used it for cross-country comparisons and database construction.

Software Implementations and Availability

The original implementation was distributed by the U.S. Census Bureau in executable and source forms compatible with platforms used by statistical offices, with code written in languages like Fortran (programming language) and interfaces for batch processing environments including Unix. Subsequent interfaces and ports appeared in statistical ecosystems connected to R (programming language), Python (programming language), and commercial systems used by institutions such as StataCorp and SAS Institute. Implementations and wrappers were developed by researchers at institutions like Carnegie Mellon University, Princeton University, and open-source contributors associated with projects hosted by repositories used by GitHub communities. Many national statistical institutes provided documentation and example workflows for operational use in production pipelines.

Criticisms and Limitations

Critics from academic centers including Columbia University, University of Oxford, and Stanford University highlighted limitations in seasonal-trend decomposition when compared with model-based alternatives like Tramo-SEATS, which was promoted by researchers at the Banco de España and scholars including Victor Gomez. Debates at conferences hosted by International Statistical Institute and workshops supported by the United Nations Statistical Division emphasized sensitivity to model selection, revisions, and endpoint behavior observed in applications at institutions such as the Federal Reserve Bank of New York and Statistics Netherlands. Limitations cited include dependence on automated ARIMA selection heuristics and challenges handling structural breaks discussed in literature from researchers at the National Bureau of Economic Research and critiques appearing in journals associated with the American Statistical Association.

Category:Time series analysis software