Economic Development News & Insight


The Higher ED Blog: Expand your analytical toolkit to get the whole economic picture

Michelle Madden / January 18, 2016

Tags: , , , ,

The Higher ED Blog: Expand your analytical toolkit to get the whole economic picture

Economic analysis is one of the core competencies of economic development, and for good reason. If we don’t understand our local economies, how can we influence their growth and development? The two most common tools used by developers are shift share analysis and location quotients (LQs). But do these tools provide a full picture?

Greg Landry, the Senior Advisor, Sector Development with the Nova Scotia Department of Business, doesn’t think so. For his Year 3 paper, Extending the Regional Analysis Toolkit, he decided to test these tools, plus multi-factor partitioning (MFP) and the Carvalho classification system, and see how the results compare. While all four are useful for understanding the performance of local industry, Greg thinks they all need to be in the toolkit because each provides a different view of economic change. To illustrate these differences, he collected Statistics Canada data for Canada’s 69 economic regions (1996 and 2012) then applied all four techniques.

The techniques

Shift Share analysis provides insight into the role of national growth and local comparative advantage in employment growth or decline of an industry. The model’s underlying principle is that a region should get its “fair share” of national growth. To determine whether a region has received its fair share or not, the model breaks down the crude growth rate of each industry into three components: reference area growth effect (the impact of national growth or decline), industrial mix effect (the performance of the industry at the national level), and differential shift effect (the impact of local comparative advantage).

Multi-factor partitioning (MFP) is a variant of shift share that has a different theoretical foundation. The creators of the model, Ray and Srinath, thought that applying the traditional model without adjusting for industrial structure would provide misleading results. Their underlying principle is that a region can only get its fair share (as determined by the traditional model) if it has an industrial structure similar to the national structure. Since each region’s structure is different, the Ray-Srinath model standardizes the data by adjusting regional industrial growth rates to reflect the relative importance of each industry to the national industrial structure. This adjustment is not the only difference; MFP has four components instead of three: regional effect, industrial mix effect, interaction effect, and allocation effect.

Location quotients (LQs) reveal the relative concentration of an industry in a region, as compared to the reference area (national) concentration. It is a simple ratio calculation where results greater than one indicate a local specialization, and results less than one indicate that the industry is lagging locally.

The Calvalho classification system uses the LQ, industrial effect, and differential shift effect results to descriptively categorize industries. The categories range from “driving” to “marginal”, as shown in Table 1. It is a useful tool for describing the current state of a local industry in a simple way.

Table 1: Carvalho Classification System

Carvalho classification

The results

The results from the various analysis techniques were usually in the same ball park, but sometimes quite different.

In a comparison of the shift share variants, the traditional shift share analysis revealed 25 regions with a positive differential shift effect for the resource and energy sector; meanwhile, the MFP analysis showed 31 regions with a positive interaction effect for the same industry.  However, the interaction effect is not necessarily a softer measure of growth as Red Deer, for example, had a positive differential shift effect but negative interaction effect. Similar oddities appeared in other sectors, like knowledge-based services. The Vancouver Island and Coast region was among those with a negative differential shift, but positive interaction effect. Calgary had a positive differential shift but negative interaction effect.

When comparing the Carvalho classification and MFP results, seven regions had “challenging” manufacturing sectors but six of those had a positive interaction effect. In the resource and energy sector, many regions were classified as “evolving”, a lukewarm category in the Carvalho spectrum, but had a strongly positive interaction effect. There were also inconsistencies between the Carvalho classification and shift share results. In the knowledge-based service sector, Toronto, Ottawa, and the Outaouais and Capitale-Nationale regions earned the highest Carvalho classification (“driving”), while Toronto, Calgary, Ottawa, and Kitchener-Waterloo-Barrie earned the highest positive differential shift effects. In case you’re curious, Toronto, Montreal, and Ottawa had the highest positive interaction effects.

More information is better

What the results above tell us is that different analysis techniques take the same raw data and find different conclusions, each with its own advantages. Traditional shift share reveals unique regional advantages, but puts more emphasis on industrial mix effect. MFP provides additional detail by dividing growth rates into finer components and taking into account the performance and concentration of other industries in the region. Location quotients and Carvalho classifications provide simple results that everyone can understand

The results of one technique are not necessarily more accurate, just different because they are based on unique assumptions. Consequently, Greg does not promote one over another, instead advocating for a more robust analysis process using all of them. When using a mix of techniques reveals “inconsistencies”, it’s a sign that the industry should get a more thorough look. Maybe it’s the academic in me, but I firmly believe that more information is better.


About the authors

Michelle Madden is the editor of Higher ED. She is also the Outreach Manager for the Economic Development Program and a graduate of the LED master’s program.  She has authored a number of the articles in this series on behalf of the students, and has published several of her own blogs on as well. Follow her on Twitter at @michelle_mad.

Greg Landry is the Senior Advisor, Sector Development with the Nova Scotia Department of Business.  He has worked in both land use planning and economic development for various levels of government.  He holds a Masters of Development Economics and a Masters of Urban and Rural Planning from Dalhousie University.

About the series

Higher ED: Insights for the Next Economy is a platform for students, guest speakers, staff and faculty of the University of Waterloo’s professional and graduate economic development programs to share knowledge with the field at large. The series takes works destined for an academic audience and reworks them into a fresh, easy-to-digest blog article.

Established in 1988, the Local Economic Development program is the only master’s program in Canada devoted solely to local economic development. It offers a balance between theory and practice by combining coursework, a major research paper, an internship, and weekly seminars featuring guest speakers. Students are prepared for careers in local, community, or regional economic development.

The Economic Development Program is a nationally-accredited provider of professional training. It delivers certification programs and seminars that offer a deep understanding of the Canadian context in a convenient block format. Peer learning is combined with informative lectures and practical case studies to provide dynamic instruction that is beneficial for junior and senior-level practitioners.

Comments are closed.