Advanced LCA – 12-716

Lecture 3

Admin Issues

• Group Projects or Take-Home Final? Your choice (individual choice)

• EIO-LCA MATLAB version - some slight improvements coming.

• HW 1 discussion

Today’s lecture

• Data sources and issues for EIOLCA

• Data consistency checks

Sectoral Classification Schemes

• ISIC – International Standard Industrial Classification

• NAICS – North Amer. Classification System

• NACE – Statistical Classification of Economic Activities in the EC

• Point: There’s a lot. For bridges between systems and comparisons, see Eurostat’s RAMON system (Google it)

History of SIC, NAICS• IO models ‘sector based’ (but have their own -

different - classification!) • Standard Industry Classification (SIC) - originally

developed in 1930s– Structures economy for data/comparative purposes– Since 30s, significant econ. changes - last updated ‘87

• North American Industrial Classification System (NAICS) - made in 1990s by US, CA, MX– Production-process based classification (similar groups)– Standard categories, country-specific adjustments– Maintains ability to compare across countries– Is in alignment with UN ISIC standard

NAICS Industry Sectors

• 6-digit NAICS codes (vs. 4-digit SIC)• First 5-digits fixed, 6th for country specifics• Example:• 33 Manufacturing [Industry Sector]• 334 Computer and Electronic [Industry Subsector]• 3346 Manufacture/Reproduction [Industry Group]• 33461 Manufacture/Reproduction [Industry]• 334612 Pre-recorded Computer CDs [Country-specific]

SIC vs. NAICS - High Level

• Agriculture, Forestry, Fishing• Mining• Construction• Manufacturing• Transport/Infrastructure• Wholesale Trade• Retail Trade• Financial/Business Services• Other Services• Public Admin (Gov’t)

• 11 Agric., Forestry, Fishing, Hunting • 21 Mining / 22 Utilities/ 23

Construction • 31-33    Manufacturing • 42 Wholesale Trade/ 44-45 Retail• 48-49 Transportation / Warehousing• 51 Information• 52 Finance and Insurance • 53 Real Estate and Rental • 54 Professional, Technical Services• 55 Management of Companies • 56 Admin, Support, Waste

Management & Remediation Services • 61 Education Services • 62 Health Care and Social Assistance

71 Arts, Entertainment, and Recreation 72 Accommodation and Food Services 81 Other Services

• 92 Public Administration

IO Model Organization

• 1997 benchmark IO tables organized into about 500 sectors

• Many IO sectors 1:1 with 5-digit NAICS• Others are 1:1 with 2, 3, or 4-digit NAICS• Others are 10:1 - e.g. agriculture• This can get really confusing!• On EIO-LCA, see “About the Model-> Sectors

in EIO-LCA

Notes on Mappings• “More high level sectors” does not alone mean “better

data” - just a different model!• Most environmental/resource data is still given in SIC

format (not yet NAICS)• Thus need multiple mapping functions• Use of (re)-mapping functions leads to additional

data/model uncertainties - hard to quantify• Auxiliaries - offices classified by ‘what they do’ rather

than ‘who they serve’– Corporate headquarters have their own sector– These offices not considered with ‘their sector’

Sample Data Mappings• For electricity consumption of some electricity

sectors, data from MECS (DOE)1

– NAICS mapping -> IO sector (easy!)

• Other manufacturing data comes in SIC– SIC -> NAICS -> IO sector (harder)

• Some no longer provided, rely on old model– Old IO -> SIC -> NAICS -> New IO sector

• Repeat 500 times (for all sectors)

1: Manufacturing Energy Consumption Survey

Old vs. New Example1992 Benchmark IO Model

Sector Economic($mill)Total for all sectors 1.671098Electric services (utilities) 1.007134Coal 0.102573Repair / maint. constr. 0.087334Crude petrol. / nat’l gas 0.041535Natural gas distribution0.037961Railroads & rail services 0.032541Wholesale trade 0.024300Petroleum refining 0.023055Real estate mgmt. 0.021044Banking 0.017472

1997 Benchmark IO ModelSector

Economic($mill)Total for all sectors 1.708177Power generation / supply 1.007417Oil and gas extraction 0.093182Coal mining 0.073502Pipeline transportation 0.031778Rail transportation 0.029385Wholesale trade 0.024219Maint. & repair constr. 0.022235Petroleum refineries 0.022115Lessors intangible assets 0.021955Real estate 0.019175

Conclusions

• Change in basis (and new data) requires considerable conversion efforts– Roughly 1000 hours to date this year

• Payoff is more up-to-date estimates of economic and sustainability metrics

• New NAICS basis should increase power for international comparisons

EIO-LCA Data Example

• Hopefully read documentation excerpt• Where does EIO-LCA data come from?• What all needs to be done to make it

useable on the web?

NAICS to IO Mapping

• Many 1:1, some complex– This comes directly from BEA

• Part 2: SIC->NAICS also from Census– http://www.census.gov/epcd/ec97brdg/– Drilldown:– http://www.census.gov/epcd/ec97brdg/E97B1311.HTM

– See “6% of, 94% of” notes..

SIC-NAICS-IO Bridge

• See documentation for details– Done in Access– Also shown in Excel (see web page)

Commodities and Industries

• Use Table basis• Commodities are produced by

industries• We have “data on industries”• Sometimes causes classification

problems• Electricity example

Electricity: As Commodity and Industry

• We have an industry called “Power generation and supply”– But the commodity “electricity” is produced by

several sectors: power gen, fed utils, state utils• We use industry by industry A matrix to better

match the industry data we have• Downside: we are modeling average

production from the industry, not “of the commodity”

• There are commodity-commodity models

Stability of IO Data

• Can see Chapter X of EIO-LCA book for more detail if needed.

• Interactive BEA IO Data Site:– http://www.bea.gov/industry/iotables/prod/table_list.cfm?anon=732

• Recall, 3 levels of detail in US IO data– Sector (~12), Summary (~90), Detail (500)

• Annual and benchmark tables– Benchmarks only every 5 yrs (next 2002!)

Stability in Sector table• Coefficients for Agriculture-Agriculture

– Direct requirements matrix values– 2005 backward to 1997– 2005 - 0.229– 2004 - 0.227– 2003 - 0.233– 2002 - 0.236– 2001 - 0.230 … (only 0.4% different)

• Can see similar stability in other cells

Allocating Inventory Data

• What happens if we don’t have source data at the “491 sector detail” level?– Running model at 12, 100, 500 level?– Compare effects of $1M production– Similar to plastic problem on HW 3/4

Case Study: Agriculture

• Mostly aggregate data.• Implications of using the data at 500

sector level• http://www.usda.gov/oce/global_change/inventory_1990_2001/

USDA%20GHG%20Inventory%20Chapter%205.pdf

• 1122 trillion BTU of energy used.• Effects on modeling?• Ideal way to deal with / show effects?