Using RegData to Answer Questions About Regulation
Download as PPTX, PDF0 likes483 views
The document discusses the impact of federal regulatory accumulation on innovation and economic growth, highlighting significant GDP losses due to increased regulations since 1949. It provides statistical analysis of regulations affecting various industries and educational spending, as well as methodologies for classifying and measuring regulatory texts. Additionally, it introduces tools such as RegData for examining trends and the effects of regulations across different sectors.
Using RegData to Answer Questions About Regulation
- 1. RegData April 8, 2015 Regulation University Patrick A. McLaughlin
- 4. Why do we need to understand regulatory accumulation?
- 5. Less innovation = less growth $15.1 trillion $53.9 trillion 0 10 20 30 40 50 60 Actual GDP Potential GDP intrillionsofrealdollars What would GDP have been in 2011 if regulations had stayed at 1949 levels?
- 6. $0 $5,000 $10,000 $15,000 $20,000 $25,000 $30,000 $35,000 1949 1951 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 inbillionsofdollars Actual and Alternative Growth Paths Hypothetical Economy Actual Economy
- 8. 0 600 1,200 1,800 0 30 60 90 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 Spending for Elementary and Secondary Schools, 1960–2011billionsofreal2013dollars real2013dollars total outlays (left axis) outlays per student (right axis)
- 9. 0 400 800 1200 1997 2000 2003 2006 2009 2012 industrymentionsintheCFR Source: Regdata 2.0. regdata.mercatus.org. Accessed August 26, 2014. Data note: Industry mentions measure the frequency with which the relevant industry is targeted by federal regulations. Industry Mentions of Elementary and Secondary Education in the Code of Federal Regulations
- 10. 0 100 200 300 400 500 1978 1982 1986 1990 1992 1994 1996 1999 2004 2008 2012 Average Math Scores (out of 500) age 17 age 13 age 9 0 100 200 300 400 500 1980 1984 1988 1990 1992 1994 1996 1999 2004 2008 2012 Average Reading Scores (out of 500) Age 17 Age 13 Age 9 Source: National Center for Education Statistics, "NAEP Long-Term Trend Assessments."
- 12. Without reading for 3 years, how can I know a regulation applies to me? WWGD
- 13. Two methods of classifying CFR text 1. Human-assisted algorithm 2. Machine learning algorithms
- 14. Two methods of classifying CFR parts 1. Human-assisted algorithm 2. Machine learning algorithms
- 15. Method 1: Human-assisted algorithm 1. Devise search terms for each sector of the economy 2. Classify documents based on frequency of search terms
- 16. Top words in Title 30: Mineral Resources Word Hits Rank shall 17692 1 be 17348 2 is 7650 3 are 3990 4 mining 3784 5 coal 3705 6 surface 3423 7 mine 3387 8 requirements 3294 9 state 3195 10
- 17. Sample NAICS 3-digit Classifications 3-Digit Code Description 111 Crop Production 211 Oil and Gas Extraction 212 Mining (except Oil and Gas) 221 Utilities 236 Construction of Buildings 315 Apparel Manufacturing 316 Leather and Allied Product Manufacturing 321 Wood Product Manufacturing 322 Paper Manufacturing 334 Computer and Electronic Product Manufacturing
- 18. Permutation algorithm to create search terms Paper Manufacturing Paper Manufacturer Paper Manufacturers
- 19. Most Mentioned 2-Digit Industries in 2012 2-Digit NAICS Classification Search Term Count Mining, Quarrying, and Oil and Gas Extraction 45,083 Transportation and Warehousing 44,712 Manufacturing 44,452 Finance and Insurance 28,683 Agriculture, Forestry, Fishing and Hunting 27,301 Real Estate and Rental and Leasing 26,282 Construction 22,261 Utilities 14,120 Educational Services 10,355 Retail Trade 4,311
- 20. Two methods of classifying CFR parts 1. Human-assisted algorithm 2. Machine learning algorithms
- 21. Two methods of classifying CFR parts 1. Human-assisted algorithm 2. Machine learning algorithms
- 22. Two methods of classifying CFR parts 1. Human-assisted algorithm 2. Machine learning algorithms
- 24. Average year: 8,100 parts 1. Train program to classify parts 2. Use training documents to teach what to look for 3. Assign probabilities for each industry-part-year
- 26. -hospitals (NAICS 622) -ambulatory health care services (NAICS 621) Associate each document with industries
- 27. Train the program to look for similar documents 1. Each CFR unit receives a probability for each industry 2. Treated as the probability the part is targeting the industry
- 28. 𝑟𝑖𝑠𝑘 = 𝑝𝑟𝑜𝑏𝑎𝑏𝑖𝑙𝑖𝑡𝑦 ∗ 𝑠𝑒𝑣𝑒𝑟𝑖𝑡𝑦
- 29. How to measure severity? • Various options – Multiply probability by restrictions – Multiply probability by word counts – Assume value of one
- 30. Weighted search term counts crop production (code 111); animal production (code 112); mining (code 212); and gasoline stations (code 447
- 32. Search term counts for finance
- 33. Summing • Each part has a number for each industry – Search terms – Search terms * severity (restrictions or words) – Machine learning-based probability * severity • Sum that industry-specific number across all parts in each year • Alternatively, select a set of agencies
- 34. Railroads
- 35. Statistics in RegData 2.0 • Restrictions • Words • Search terms (2-, 3-, and 4-digit NAICS) • Regulation index (search terms * restrictions) • Years 1997 – 2012 • Agency-specific
- 36. Coming in RegData 2.1 • Machine learning-based probabilities • Each part associated with underlying statute • Years 1975 - 2014
- 37. What you can do with RegData • Examine agency trends • Examine industry trends • Combine agencies and industries
- 38. Dept. of Labor – Word Count
- 39. OSHA – Restriction Count
- 40. Mining, Quarrying, Oil, and Gas
- 41. Mining, Quarrying, Oil, and Gas by EPA
- 43. Railroads by FRA and PHMSA – Search Term Count
- 45. Dept. of Treasury – Word Count
- 46. Treasury Breakdown - Customs
- 47. What you can do with RegData • Examine agency trends • Examine industry trends • Combine agencies and industries