This set of web pages is related to Andersen's project on the use of phylogenetic/phenetic methods in evolutionary economics. It contains information, data and links on the reconstruction of industrial similarity trees and/or genealogical trees. This reconstruction is done by means of the methods for finding a tree with a good fit to the data within a reasonable time limit that are provided by computer scientists and biological researchers. A special emphasis of put on the phylogenetic methods as they are developed by molecular biologists. A few links to other researchers in the field are given in the database with links to evolutionary economics et al.
Data of industrial characteristics formatted for phylogenetic analysis
Several datasets are made available to promote the exploration of phylogenetic (or cladistic) methods. The datasets are primarily intended to be used with the package MEGA2, but they are also available as simple tables that can be imported into e.g. Excel or MacClade.
- Data on the characteristics of apes (mitochondrial DNA), from GenBank - cf. Kumar and Nei (2000), pp. 90-92.
- Data on the characteristics of automobile assembly organisations, cf. the literature on the cladistics approach.
- Data on the characteristics of industries, based on the OECD/STAN input-output database.
US input-output characteristics for 23 industries (CIIO)
- Initial experiments are performed by means of industrial characteristics derived from input-output tables. A nice set of raw tables can presently be downloaded freely (in Lotus wk1 format) from the OECD/STAN input-output tables.
- The characteristics have been converted for use by phylogenetic software. To give a quick impression there is a table with example data and some related discussion. Several characteristics sequences for an individual industry i have sometimes been concatenated. If we call the input charecteristics sequence CI[i] and CO[i] is the sequence of output characteristics, then CIO[i] is the concatenated sequences. Similarly we construct CIIO[i], etc. It is also possible to combine sequences for different years and different countries, like CIO[US72,i]xCIO[US90,i] or CIO[US72,i]xCIO[UK68,i].
The reconstruction of industrial trees is performed by means of phylogenetics software. Since this software comes from biological taxonomy and molecular biology (see core references), some care is needed. The industrial characteristics data must be supplied in an appropriate format (see above). An understanding of the underlying algorithms and the relevant options and features of the software is important.
Example results produced with MEGA2
- Do you want to produce one of the following results yourself?
- Download and install MEGA2, download one of the MEGA2 data files, run MEGA2, activate the data file, change to Distances/Choose Model/Nucleotides/No. of differences (or p-distances), and choose the algorithm froim the Phylogeny menu. Then you get one of the following trees.
- Naive phylograms of apes by the minimum evolution method - using the simple number of differences between the sequences. Based on mitochondrial DNA.
(a) The basic structure of the tree
(b) The distances between the species and their reconstructed ancestors
- Phylogram/cladogram of automobile assembly organisations by the maximum parsimony method method. Based on 44 automobile assembly characteristics.
- Naive phylogram/phenogram of industries based on a limited number of industrial characteristics. From input-output data.
Example tree by the cluster analysis method (UPGMA) from the CIIO characteristics data for 23 industries for US 1990.
Comment: The tree demonstrates that no simple interpretation is possible. The poorness of the data (only 2 x 35 characteristics) and the inclusion of industries at several levels of aggregation bring surprising industries together (like food and drugs). The project tries to find better data that reveals both deeper similarities and evolutionary history.
- The OECD/STAN version of the ISIC classification of industries (drawn by MacClade)
The above example industry tree may be compared with the tree below that depicts the hierarchy that is implicit in the ISIC classification system (see the page on the classification).