Economic Geography

The research question(s) frame the methods needed
Most research projects that I have undertaken have been motivated by either my
own curiosity or by an outside request. In both cases the research questions have
generally been ‘on the table’, and have defined how I have approached my research.
In some cases my research has been stimulated by the work of others, and I have
chosen to push the envelope on that work, and to develop these themes further.
I will use three papers to illustrate this approach.
Washington State was a pioneer in the development of survey-based input–
output tables, producing the first table benchmarked against the year 1963, as
well as tables for 1967, 1972, and 1982. One of my first papers used methods
developed by Leontief and Carter to analyse structural change in the 1963
and 1967 Washington input–output models (Carter 1970; Leontief 1953). This
approach required standardizing the definitions of the sectors for both years,
then calculating inverse matrices, and then properly multiplying final demand
vectors to estimate output. The mathematics involved was exactly as developed
by Leontief and Carter, although we did not have the resources to engage in the
types of price standardization that characterized their research. At the time that
this research was undertaken, we did not have evidence regarding the stability of
multipliers in regional input–output models. The paper found that the regional
structure was less stable than the technical requirements, but also found evidence
of business cycle effects (Beyers 1972). Further research of a similar nature by
Conway also found business cycle effects, and helped make the case for periodic
re-measurement of regional input–output relations (Conway Jr 1977). Many
years later I revisited this topic, with data available over a much longer time
period, using data from nonsurvey updates of the Washington input–output
models, and I found that changes in regional interindustry structure had been
modest, even though the shares of output of various sectors had been dramati-
cally altered (Beyers 2001).
Another paper involving input–output models made use of existing models
and research methods developed by others to focus on the empirical identifica-
tion of key sectors. However, in this chapter I innovated the use of input–output
multipliers for the analysis of forward linkages using an inverse matrix based on
sales coefficients, and the use of purchases coefficients to derive multipliers related
to backward linkages (Beyers 1976). The paper also cast these measurements into
an interregional model environment, and showed how change in geographic scale
influenced the identification of key sectors.
A third paper (with David Lindahl) used Michael Porter’s definitions of compet-
itive strategies to classify responses of a set of producer service establishments to

---

certain questions regarding firm’s perceptions of the bases of their competitive
edge. We then tested the performance of these firms in terms of sales per
employee and growth in sales, to ascertain which of these competitive strategies
were superior for producer service establishments (Lindahl and Beyers 1999). In
this paper we tried to be faithful to Porter’s definitions of competitive strategies,
but we also used discriminant analysis to demonstrate that there were other viable
strategies being employed by producer services beyond those defined by Porter.
Each of these papers were driven primarily by methods or models developed
by others, and the primary goal was to provide evidence in a different environment
of their robustness.