From the Orderly World of Frameworks to the Messy World of Data: Canadaâ•Žs Experience Measuring the Economic Contribution of Maritime Industries

This paper extends the frameworks developed by Park and Kildow (2014) to explore the use of supply chains to classify and organize ocean industries. Canada’s experience with measuring the economic contribution of ocean related sectors is discussed; with particular emphasis on illustrating the supply chain approach and of highlighting the multiple practical challenges that arise in defining and measuring the ocean economy. The overall conclusion is that Canada has succeeded in developing a methodological framework that allows it to report, on an annual basis high level estimates of gross domestic product, employment and labour income generated directly and indirectly by ocean sectors in Canada. The main challenge encountered is the potential for double counting, as a result of poor industry definition and use of input-output models to calculate total impacts (i.e. direct, indirect and induced). This research article is available in Journal of Ocean and Coastal Economics: https://cbe.miis.edu/joce/vol2/iss2/9


INTRODUCTION
The Canadian experience measuring the economic contribution of ocean industries 1 started in the late 90's, when the first report "Canada's ocean industries: contribution to the economy, 1988-1996" was prepared by Roger A. Stacey Consultants Ltd. (1998. This report provided the first compilation of ocean related industries in Canada and offered a first appreciation for the challenges of gathering appropriate data for estimating the economic contribution of these industries (GSGislason, 2007). The work was updated in 2003 to encompass estimates for 1988-2000.
The Roger A. Stacey Consultants Ltd. reports covered the largest maritime industries in Canada: seafood, offshore oil and gas, ocean transport, ocean tourism, marine construction, ocean manufacturing, and government. Despite the good coverage, some gaps were pointed out by GSGislason (2007), such as university and research related expenditures in the public sector, ferry revenues in ocean transport, and self-guided tourism and recreational activities in ocean tourism. In addition, it must be noted that provincial government expenditures, and support activities to offshore oil and gas and marine transportation were not included in the report.
A further limitation in the scope of this first reporting effort was that only direct impacts were estimated, which left out spill over (indirect) impacts. The study did not use Statistics Canada's Interprovincial Input-Output (IO) model, which is available since 1961 on a national basis and since 1997 at the provincial level 2 . The IO model is the most comprehensive articulation of economic activities and flows of goods and services in the Canadian economy.
Seeking to build on the work by Roger A. Stacey, and aiming to develop a national framework that captured all relevant industries and allowed for estimating the multiple layers of economic contribution of ocean industries, Fisheries and Oceans Canada first commissioned GSGislason to prepare a methodology for reporting on a marine sector national report card (2007), and subsequently retained 1 The terms "ocean", "marine" and "maritime" industries are used interchangeably in this report.
Gardner Pinfold to further refine the methodology and develop the national report card (2009). The outcome was a framework that estimated the direct, indirect and induced economic contribution of a rather comprehensive set of ocean industries in Canada, with a clearly articulated methodology that addressed the most important concepts utilized as well as the limitations of the data and methods employed. This report marked the first time that Statistics Canada's IO model was used to estimate the economic contribution of maritime industries in Canada. The resulting report was peer reviewed by a number of international and Canadian experts.
Subsequent efforts have since focused on developing a time series of the economic contribution of maritime industries in Canada for use in policy development and analysis, and on assessing the feasibility of extending the coverage to ocean related activities in Canada's Arctic.
Canada has adopted a definition of ocean economy mainly focused on the natural resources of the ocean (Colgan, 2003). Ocean industries have thus been identified based on their use or exploitation of ocean resources or their linkage to industries that do so, rather than on their location along coastal areas. Exceptions to this are the marine tourism and recreation sector and the universities sector. The former includes a coastal focus in that it considers that some activities that take place along the coast are related to the enjoyment of the ocean (e.g. national parks located along the coast, visits by tourists to coastal towns). University related expenditures are included for coastal universities, which may have left out some research centers located away from the coast.
Further, the scope of Canada's ocean resources considered is geographically delimited within the country's exclusive economic zone (EEZ), and the scope of industries included encompasses businesses that operate in Canada. As a result, the use or exploitation of non-Canadian (foreign) ocean resources by domestic industries or firms is excluded (e.g. Cooke Aquaculture, a Canadian owned company, has aquaculture operations in other countries that are not included in Canada's ocean economy). Similarly, the use or exploitation of Canadian ocean resources by foreign companies not based in the country is also excluded (e.g. foreign owned cruise ships operating in Canadian waters, whose revenues do not stay in the country, except for passenger expenditures in the port of call; or foreign based submarine cable companies that use Canada's ocean floor, which may pay the government for permits but do not bring revenues to the country).
It is important to note that Canada's efforts have so far focused exclusively on market related activities 3 , as these are most easily measured (Colgan, 2003). The non-market economy has not yet been measured, although there are ongoing efforts to assess the economic value of the subsistence and barter economy in Canada's Arctic region. In addition, the government of Canada has undertaken some efforts towards measuring ecosystems goods and services, including those provided by marine ecosystems 4 .

Classification of Industries
A lot of effort has been undertaken by various researchers and countries over the past decades in defining and measuring the ocean economy. Park and Kildow (2014) conducted a comprehensive overview of the literature on the ocean economy and of the studies carried out by various countries in this regard. In their paper, they propose two very useful classifications for scoping and organizing ocean industries, as well as an international standard of ocean sectors that could be applied to the reality of just about any country for the purpose of facilitating international comparisons or aggregations.
Park and Kildow (2014) use two different perspectives for scoping and organizing ocean industries. The first classification is based on the relationship of the industry to the ocean resource or to other industries that use the ocean resource. Industries can hence be classified in three groups: "in the ocean", "from the ocean" and "to the ocean". "In the ocean" industries are those that directly use, protect, research and develop the ocean (e.g. fish harvesting, marine shipping, offshore oil and gas). "To the ocean" industries are those that supply inputs to the first ones (e.g. ship/boat building, marine manufacturing and construction, support services to marine industries), and "from the ocean" industries are those that add value to the outputs of the first ones (e.g. seafood processing, petroleum refining, marine biotechnology). The second classification uses the supply chain approach, focusing on the supply chain relationship among various industries that utilize an ocean resource. Industry clusters can thus be formed around ocean resources by linking the industries that directly use or harvest the resource with those that are downstream (i.e. add value to the ocean resource) or upstream (i.e. supply inputs to the "direct" industries).
Both classifications are compatible and can be readily combined. Taking commercial fish resources as an example, fish harvesting takes place "in the ocean", while fish and seafood processing and fish distribution/wholesale/retail use the resource "from the ocean" (fish) and add commercial value to it; in turn, ship yards, fuel stations and fishing gear manufacturers amongst many others, supply inputs "to the ocean" industries that directly use the resource (fish harvesting). An industry cluster is hence built around commercial fish resources composed of many industries: ship building, fuel stations, textile product mills, fishing, seafood processing, and seafood wholesale and retail, to name a few.
These classifications offer a framework that in the case of Canada can be used in conjunction with the North American Industrial Classification System 5 (NAICS) to scope out the economic industries to be included in Canada's ocean economy. An illustration of this is offered in Figure 1 (next page), where commercial fish resources are used to illustrate the industries that contribute economic value along the supply chain (as per previous paragraph's example).
5 http://www.statcan.gc.ca/pub/12-501-x/12-501-x2012001-eng.pdf The NAICS provides a practical framework for implementing the conceptual definitions of maritime industries. The key advantage of this system is that it is used by Canada, the United States and Mexico as industry classification standard. Canada's national statistics agency (Statistics Canada) uses it for reporting on industry statistics and for developing the country's input-output model (see Table  1 on following page), which is the primary modeling tool used for estimating the economic contribution of maritime industries in Canada. In addition, the NAICS meet all objectives proposed by Colgan (2003). It must be noted that in addition to industries that are linked commercially, the supply chain/NAICS framework has been extended to include public and civil sector organizations (government departments, universities, social advocacy organizations). These sectors are commonly mandated with or vested in the stewardship or management of commercial fish resources 6 , and are therefore concerned with how, in what manner and to what extent the commercial fish resources are harvested.
In fulfilling their role, these sectors undertake activities with the goal of generating knowledge, managing the resource and providing stewardship to the exploitation of commercial fish resources. Hence, these sectors do not sell goods and services to the fishing industry; instead they contribute to the economic value through the generation of knowledge, management of the resource and provision of stewardship.
Canada's ocean industries data encompasses the majority of industries that form the commercial fisheries cluster (blue color bubbles in Figure 1). However, the economic contribution of seafood wholesale and retail are not included (red color bubbles in Figure 1). Since most of Canada's fish and seafood production is exported to international markets, with the corresponding economic value "leaking" out of the Canadian economy 7 , the omission of these two industries likely results in a rather small underestimation of the economic contribution of this cluster 8 . Similar clusters can be built for other industries, with a similar supply chain flow. The offshore oil and gas cluster ( Figure 2) shows upstream linkages to industries that supply engineering services, support activities and boats/vessels among others, and downstream linkages to industries such as pipelines and refining, chemicals manufacturing, natural gas distribution and wholesale/retail of fuel products. The marine transportation cluster ( Figure 3) portrays an industry that provides what could be considered a final service, in the sense that there are not many downstream activities or industries that could add value to marine shipping 9 . This might reflect the fact that there is no tangible good extracted from the ocean upon which further processing or value added can be applied.  Based on these industry classifications and their review of country specific reports, Park and Kildow (2014) developed a proposed international standard for the ocean economy that consists of 12 sectors (Table 1). Canada's ocean industries data covers eight out of these 12 sectors, which in the case of Canada are the largest ones. The sectors or industries that are less well represented are marine mining, marine equipment manufacturing, marine business services, and other (mostly emerging) industries. However, some of the economic activity associated with these sectors is likely captured when estimating indirect impacts through the use of input-output models. As Colgan (2003) suggests, the economic activity associated with secondary and tertiary sectors with intermediate connections to primary industries (such as marine manufacturing and business service industries) can be best estimated using national input/output tables.

Commodities Based Activities
Despite the development of a Tourism Satellite Account and National Tourism Indicators 10 , the Canadian marine tourism and recreation sector remains a collection of industries that are independently classified under the NAICS. Many of these industries (e.g. restaurants, car rental) include a large portion of nontourism related activity, which exacerbates the already big challenge of teasing out the marine related share. Moreover, rather than an industry, marine tourism and recreation may be seen as a collection of activities undertaken by final consumers (tourists and recreationists). Hence, an alternative approach could be beneficial in portraying this activity.
Tourism and recreation activities are commonly measured through the amount tourists or recreationists spend on a variety of commodities (e.g. accommodation, food services, car rental, travel fares). Figure 4 shows a graphic definition of marine tourism and recreation through the use of Statistics Canada's Input-Output Commodity Codes classification (IOCC) 11 . The bubbles in this diagram represent the commodities purchased by tourists and recreationists in pursuit of tourism and recreation activities, rather than the industries that provide these commodities 12 . Gardner Pinfold (2009) collected expenditure data by commodity type according to the IOCC and used the commodities table of Statistics Canada's IO model to estimate the associated economic impacts (Table 2). IO models connect commodities to industries, so impacts can be traced to industries involved in the ocean economy.

FRAMEWORK OBJECTIVES
The US National Ocean Economics Program has developed a set of four objectives to inform the development of a framework to measure the economic contribution of ocean industries (Colgan, 2003): 1. Comparability (consistency) across industries and space 2. Comparability (consistency) across time 3. Theoretical and accounting consistency (i.e. no double counting) 4. Replicability These objectives are very similar, if not the same as the ones used by Gardner Pinfold (2009) in their analysis of the economic contribution of ocean industries in Canada. As noted by Gardner Pinfold, comparability across industries, geographies and time are greatly enhanced by classifying industries according to the NAICS and by obtaining statistical information for all industries from a unique source (Statistics Canada). However, despite any efforts in this regard, data availability and industry definitions often conspire to introduce imperfections in comparability and consistency. In the case of Canada, the following industries have presented issues with obtaining comparable data across industries and sometimes across geography (provinces): a) Offshore oil and gas: value of output data has been suppressed by Statistics Canada due to confidentiality concerns; the alternative used has been production data published by provincial petroleum boards together with average market prices quoted by the United States Energy Information Administration and exchange rates published by the Bank of Canada.
b) Marine transportation (shipping): data is also suppressed by Statistics Canada due to confidentiality concerns; the alternative used is custom statistics on industry revenues prepared by the Canada Revenue Agency. d) Shipbuilding and boat building (includes offshore oil and gas drilling and production platforms): data is suppressed for confidentiality reasons for some provinces; Statistics Canada's Business Register data on employment and establishment counts is used to prorate national estimates. e) Marine Construction: port and harbor related construction is included in a broader NAICS code (237990: Other heavy and civil engineering construction); capital expenditures on construction by type of asset published by Statistics Canada have been used as an alternative.
f) Government/Public administration: government departments, whether federal or provincial, oftentimes have mandates that overlap marine and land related roles; government public accounts have been used in conjunction with expert judgment and special requests to some government departments to discern the marine component.
g) Social advocacy organizations: income and gross domestic product (GDP) by primary area of activity for non-profit institutions and volunteering was terminated in 2008; expenditures for a representative sample of marine-related environmental non-government organizations (ENGOs) have been used instead. h) Universities: universities undertake research in a broad range of disciplines, only a subset of which is related to ocean resources; data and information on ocean related grants to coastal universities have been used in conjunction with average salaries for professors in ocean-related institutes associated with coastal universities. Meeting the objective of comparability over time has also presented challenges. The discontinuation of data sources, whether of data published by Statistics Canada or reports prepared by/for industry associations, is an ongoing difficulty, particularly with diminishing budgets in the public administration. Examples are the recreational fishing survey regularly undertaken by DFO, which is currently delayed, and the recreational boating survey, which has been discontinued. Data confidentiality can also pose a problem in this regard, as data may be confidential in some years but not in others, particularly at the provincial level.
Another big challenge related to creating time series data is the cost of gathering all value of output and expenditure data required for estimating economic impacts through an input-output model. Canada's experience suggests that benchmarking studies may be conducted approximately every five years, with ongoing annual updates based on readily available proxies. Benchmarking studies are often contracted out to consulting economists, who bring a wealth of expertise and knowledge as well as industry contacts, particularly concerning industries that are outside the mandate of DFO. Cost and expertise considerations make it difficult to replicate benchmarking studies with internal DFO resources.
The use of the NAICS and of Statistics Canada's interprovincial IO model provides a reasonable guarantee of theoretical and accounting consistency. However, the challenge of double counting has not been entirely eliminated from Canada's estimates. The risk of double counting is highest when ocean industries purchase inputs from other ocean industries (Pugh, 2008 andOxford Economics, 2013). In these cases, the value added of a sector included as a separate industry can be double counted as part of the indirect value added of other maritime sectors it supplies goods or services to. Examples are commercial fishing and seafood processing, shipbuilding and marine transportation, and support activities to marine transportation or to offshore oil and gas and their respective direct industries. In these instances, the "in the ocean" activity (commercial fishing) is double counted to some extent in the indirect impacts corresponding to the "from the ocean" activity (fish and seafood processing), or the "to the ocean" activity (shipbuilding, support activities) is double counted to some extent in the indirect impacts corresponding to the "in the ocean" activity (marine transportation, offshore oil and gas).
Canada's ocean industries estimates include these industries separately, without proper adjustment to indirect impacts to eliminate (or minimize) double counting.
It is difficult to estimate with exactitude the magnitude of double counting in Canada's estimates. However, the following may provide a general appreciation of the problem 13 : a) Support activities for offshore oil and gas are estimated to contribute CDN$208 million 14 (2012) in total economic impacts. A portion of this would be double counted in the indirect impacts of offshore oil and gas industry.
b) Support activities for marine transportation are estimated to contribute CDN$4.6 billion (2012) in total economic impacts. A portion of this 13 For context, the total value of Canada's marine industries was estimated at CDN$36.1 billion (2012): http://www.dfo-mpo.gc.ca/stats/maritime/tab/mar-tab1-eng.htm. 14 All dollar figures presented in this paper are expressed in Canadian dollars (CDN$).
would be double counted in the indirect impacts of marine transportation.
c) Ship and boat building are estimated to contribute CDN$984 million (2012) in total economic impacts. A portion of this would be double counted in the indirect impacts of marine transportation, commercial fishing, aquaculture, and marine tourism and recreation.
Gardner Pinfold (2009) made some adjustments to commercial fishing and seafood processing to avoid double counting. This correction was done by setting purchases from the fishing industry to zero when estimating the processing industry impacts. DFO is planning on addressing double counting problems more broadly in the next benchmarking study.
The replicability of estimates is affected by some of the challenges already noted, such as the discontinuation of data sources and the suppression of data due to confidentiality concerns. Hiring external consultants does tend to add to the complexity of the quantification effort itself. Private consultants often specialize or find niches, either through their accumulated knowledge and expertise or through their networks of contacts. This can make it difficult for other consultants to fully replicate their methodology. In addition, extra efforts have been required on occasion to get precision on data sources utilized or to obtain copies of materials used (e.g. spreadsheets with calculations or results).
Seeking to make estimates replicable, DFO has developed a spreadsheet based methodology for updating estimates on an annual basis. Due to unavailability of some data sources (i.e. discontinued: marine construction, or infrequent: recreational fishing) and difficulties replicating the methodology for particular sectors 15 (marine transportation), the model uses a combination of the methodology developed by the latest benchmarking study (commercial fishing, aquaculture, seafood processing, offshore oil and gas) together with proxy indicators to generate growth rates that are applied to benchmarking study results (remaining sectors). This has allowed for reasonably accurate estimates that can be used in high-level policy analysis.
15 Refer to previous paragraph for a discussion on challenges replicating estimates.

ALLOCATING INDIRECT AND INDUCED ECONOMIC IMPACT ESTIMATES
The use of input-output type models facilitates in great fashion the estimation of economic impacts at the industry and regional (provincial) level, and allows for the capturing of the value added generated by upstream industries and by labor demand (i.e. indirect and induced impacts). The results of the input-output model show the entire flow of economic activity throughout industries and provinces (regions), which can be readily used to portray the economic contribution of the ocean sectors. The linkages to land-locked regions and to land-based industries can be then directly observed. As shown in Figures 7 and 8, impacts spread beyond Canada's marine coastal provinces and beyond ocean industries.
However, there may be merit in presenting the results in such a way that all economic activity that is triggered by an ocean industry is combined to show the aggregated economic impact of the ocean industry, including impacts that arise in non-ocean sectors 16 . Likewise, economic activity may be aggregated at the provincial level to show the cumulative economic impact of any one province's ocean economy, regardless of the province where the impacts occur.
The allocation of direct economic impacts is straightforward, since they reflect the value added by the industry involved in the direct activity and take place in the province where the industry operates. Hence, direct impacts will always accrue to ocean industries and coastal, marine provinces.
Allocating indirect and induced impacts presents challenges. These straddle both coastal and in-land provinces. For example, the latest benchmarking study undertaken by DFO indicates that approximately 10% of GDP and 11% of employment generated by Canada's maritime industries occur in non-coastal (inland) provinces (Figure 8). This is an average for all industries included in the study. For some industries the percentage of economic impacts occurring in noncoastal regions is likely bigger. (1) Central provinces include Ontario, Manitoba, Saskatchewan and Alberta.
(2) Economic impacts include direct, indirect and induced.
A similar challenge arises because there are many non-maritime industries involved in supplying maritime industries, hence creating economic value that is indirectly related to the ocean economy. The same Canadian benchmarking study suggests that in 2008 approximately 32% of GDP and 45% 17 of employment were generated by industries that have little or no ocean related component (Figure 9). Thus far, DFO's approach has been to re-allocate all indirect and induced impacts to the coastal provinces where the direct activity that generated the indirect or induced impacts took place. A similar approach has been used for industries, whereby indirect and induced economic impacts are re-allocated to the maritime industries that originated or triggered the indirect economic activity. This is done through the use of multipliers, which are available at the industry level (i.e. NAICS) and at the provincial level. Alternatively, IO model results may be calculated separately for each industry and for each province, although this has a rather high monetary cost.

CONCLUSION
The contribution of this paper consists in placing the experience of quantifying the economic contribution of ocean sectors in Canada within the context of current methodological frameworks, and in reflecting on some of the challenges encountered in pursuing the objectives proposed in the literature for implementing these frameworks.
Canada's experience thus far shows that despite the inherent challenges of the task, it is possible to integrate disparate data sources into a coherent framework that provides robust estimates of the economic contribution of maritime sectors in Canada. Currently, economic impact estimates (direct, indirect and induced) are available for gross domestic product (GDP), employment and labor income by industry and by province, for the years 2006, 2008-2012 18 . All major ocean industries are included in the estimates. Benchmarking studies are conducted approximately every five years and annual updates are prepared in between.
This overview of frameworks and of Canada's experience show that Canada's maritime industries data does not include some ocean related sectors, most notably marine pipelines and refineries, marine equipment manufacturing, marine business services and emerging industries (renewable energy, undersea cables), although the economic impacts associated with these industries are likely accounted for at least in part within the indirect impacts associated with the ocean industries they supply.
18 http://www.dfo-mpo.gc.ca/stats/maritime-eng.htm Double counting seems to be the main problem affecting Canada's data. As a result, the economic contribution of some sectors may show an upward bias. The magnitude of this bias has not yet been quantified, although it is an area for future work. The main industries affected are: shipbuilding-marine transportation, support services-marine transportation, and support services-offshore oil and gas.
A potential solution for dealing with double counting would be to estimate the economic impacts at the level of ocean industry clusters (i.e. supply chain). This may require calculating the IO model results for the "in the ocean" (marine transportation) or "from the ocean" (seafood processing, retail gas stations) industries, as these are closer to the final demand and would therefore capture all upstream industries through indirect impacts (IO model). However, this presents its own problems, which could be the subject of another paper. In particular, industry classifications (NAICS) are typically broader than their ocean component. For example, retail sale of gasoline includes supplies from land-base as well as marinebased oil rigs. In addition, imports and exports are more difficult to trace or track. Continuing with the same example, the gasoline purchased at the pump could originate from marine-based oil rigs located in Canada or abroad. Therefore, careful consideration must be given before embarking in this approach.
Other challenges arise, which are common to most if not all studies reviewed, concerning suppressed data due to confidentiality issues, discontinuation of data sources, or plain unavailability of output or expenditure data for some industries. This remains a lesser challenge, and one that cannot be fully eradicated. As Pugh (2008) suggests, aggregate estimates of the economic contribution of maritime industries represent a ballpark. Canada's ocean industries data series does at a minimum meet this qualifier.
The next steps in quantifying Canada's ocean economy include the conducting of a benchmarking study for 2013 (time lag due to data availability). This study will place particular focus on expanding the geographical scope of Canada's ocean sectors data to include Canada's Arctic and on avoiding double counting. Future goals would be revisiting the definition of industries included in Canada's ocean economy in light of the clusters framework to ensure complete coverage, and seeking to enlarge the scope of the data set to include some of the new and emerging industries (renewable energy, sea bed cables).