About

 

This blog documents and discusses World-System state space models for the period (0-2000) using data from the Maddison database (for more detailed information, see the Boiler Plate). Constructing models covering such a long period of time brings up many issues: (1) Theory supporting the models, (2) Periodization, (3) data availability and comparability across time, (4) handling missing data (see the  Boiler Plate for more information) and (5) a long list of miscellaneous problems. 

Theory

The theories that support this project are World-Systems Theory, General Systems Theory, Impact Models to include the Kaya Identity, Cybernetics to include Control Theory, Unified Growth Theory, Balanced Growth Theory, State Space Theory, and Stability Theory. The purpose of invoking all these theories is to link my results with the work of the Intergovernmental Panel on Climate Change (IPCC). All the models are estimated statistically using Dynamic Components Models (DCMs) where the approximate system state is estimated first using Principal Components Analysis (PCA) and then using the Kalman Filter. The models are analyzed using Computer Simulation.

Periodization

The graphic above shows my approximate periodization of World History from (0-2000). When using statistical DCM models, there is no reason to stick to rigid periodizations of history. However, all historians use some type of explicit or implicit periodization that fits (roughly) the graphic above: the Roman Empire, the Dark Ages, the Middle Ages, Feudalism, Capitalism, British Capitalism, and the Take-off Into Unstable Growth (other periodizations are possible).

Data

There is, understandably, a shortage of data describing the World-System.  Data shortage forces the wide-spread use of Qualitative history. Historians search archival records or read secondary sources and try to reconstruct what happened during a specific historical period.  Quantitative History uses whatever data is available or can be constructed from archival records to present time plots of trends. Models are seldom estimated because of missing data. 

The database developed by Angus Maddison (1926-2010) was a major advancement in the collection and standardization of historical data. There are small pockets of other data (particularly covering the British Capitalism period and used extensively by Cliometricians**) and data sets generated by the statistical services of selected governments (for example, the Historical Statistics of the United States and the World Development Indicators). I have used all these data sources to answer particular questions. My approach to data sources and the handling of missing data is discussed more fully in the Boiler Plate.

Acknowledgments

I have already acknowledged my huge debt to Angus Maddison (1926-2010). I also have made extensive use of Wikipedia, an on-line encyclopedia, and Google searches. I have made less use of the Academic literature but, when I do, I typically made references to Jstor. My goal in providing links is to give readers background in the topics I am covering. None of these source is perfect (nothing is), and the reader should consider all historical and theoretical conclusions tentative.

My main purpose is to construct state space models. All my computer-statistical models are written in the R programming language and are available for you to run on-line as code Snippets stored on Google sites. I have used this approach in courses I have taught at the University of TN--Knoxville. I usually take the first few days in class to explain how to run the models and how to make changes to the R-code for students to experiment with the models and document what they have discovered in Reaction papers. At the end of the term, I have them work in teams to give presentations on what they have learned.

Notes

** Cliometricians have typically used economic models to generate all output data and compared the results with whatever historical data was available. Parameter values in models were typically "calibrated" on a single year (see Williamson, 1974).