Lessons in Data and Statistical Analysis. NCSL Fiscal Analysts Seminar October 5, 2017

Transcription

1 Lessons in Data and Statistical Analysis NCSL Fiscal Analysts Seminar October 5, 2017

2 Topics for Consideration: Data types, relevance and reliability Useful data sources Tools Examples Pitfalls and Problems

3 Evaluating reliability and relevancy Source Bias Tips for assessing the credibility of online information (Center for Public Education, based on Stanford Guidelines for Web Credibility) 1. Does the Web site provide references for research that can be independently verified? 2. Are authors identified? Are their affiliations, credentials, and contact information provided? 3. Who owns or is responsible for the Web site? Is a physical address and complete contact information provided? 4. Does the site describe its mission? Are staff members identified? 5. Does the site carry advertising? If it is run by a not-for-profit organization, are its sources of funding identified? 6. Is the site professional in appearance and quality? How recently has it been updated? Is it free from typographical and grammatical errors?

4 Evaluating reliability and relevancy Time Period

5 Evaluating reliability and relevancy Unit of Analysis individuals households

6 Evaluating reliability and relevancy Common Adjustments Inflation (what year = 100?) Seasonally-Adjusted (SA) Seasonally-Adjusted at an Annual Rate (SAAR) Per Capita

7 Evaluating reliability and relevancy Flaws/Errors Statewide In-Migration vs Utah School Age Enrollment Statewide In-Migration = 11,000 School Age Enrollment Growth = 12,161 (School Migration Est. 5,355) Under estimate = 4,243 Cost = $17 million

8 Useful data sources Bureau of Economic Analysis Consumer Expenditure Survey Bureau of Labor Statistics U.S. Census Bureau National Bureau of Economic Analysis American Economic Association State Vital Statistics Offices Global Insight Moody s Federal Reserve

9 Data Sources: BLS Consumer Expenditure Survey (CES) Cross-section of spending characteristics of US households Useful for analysis of sales tax proposals for narrow taxes/ exemptions: Exemption bill on diapers/ feminine hygiene products Bill expanding sales tax to services

10 Data Sources: FRED Federal Reserve Economic Data (time series) BEA, BLS, Census, numerous other data releases Always up-to-date on latest data releases Example: LA Sales Tax Able to manipulate data as needed

11

12

13 Data Sources: Other Agencies Federal Reserve Banks and Board of Governors Continuously research policy issues and monitor economic trends Timely research on net impacts of sales tax holidays Crucial for analysis of LA Second Amendment Sales Tax Holiday bill

14 Data Sources: Other Agencies US Energy Information Administration (EIA) Continuous research and monitoring Timely research on impact of gasoline prices on consumer response Useful in analysis of LA Gasoline Tax increase bill

15 Data Sources: Other Agencies US Department of Agriculture Extensive data on agricultural production Current data on farm size and production by state Vastly improved estimate reliability on proposal exempting irrigation materials from sales tax

16 Data Sources: Market Research Firms Example: Statista Limited amount of free data, but enough US-level info to be useful Examples: Expansion of sales tax to services including digital media Diapers/feminine hygiene products sales tax exemption

17 Other Data Sources Professional or Industry Associations Often can be a valuable source of very relevant data (but be aware of their mission/motives) Example AOPA: Aircraft painting service tax exemption Think Tanks or Professional Research Groups Sometimes can function similarly to an industry association; same caveats apply Example University of New Orleans Hospitality Research Center: Louisiana Tax Free Shopping Program for foreign tourists

18 Tools Analytical Tools Excel Application Pros Cons Ubiquitous Easy to present results on the fly Hard to track iterative process Size limits (better) Statistical Packages (R, STATA, SPSS, SAS, EVIEWS, ) Intuitive Good documentation of process Sophisticated modeling techniques Large datasets/faster Can be slow Steeper learning curves Expensive Output presentation can be tricky

19 Statistical Packages: STATA Example freduse

20 Example Utah Tax Reform Bill of 2006 Modified Flat Tax Went from 6 bracket to one with deductions being converted to credits. Tax Commission had access to the filer data but LFA and Governor did not. Tax converted the data into an Excel spreadsheet. After the bill passed the Tax Commission discovered an error in the spreadsheet understating the revenue loss by $200 million. Resulted in a change in statute to allow LFA to access taxpayer data as long as the taxpayer could not be identified. Transitioned from Excel to SAS for analysis.

21 Taxpayer Receipt Tools budget.utah.gov Environment and Natural Resources Economic 3% Development 2% Criminal Justice 7% Infrastructure 10% Social Services 14% Elected OfficialsGeneral 1% Government 0% Public Education 48% Higher Education 15% State Sales and Income Tax Uses Enter your income tax paid: $15,000 Enter your sales tax paid (or "x" to estimate from income tax): x Total: $21,100 $6,100

22 Taxpayer Receipt Statewide Agency (in M) Your Share Grand Total $ 6,266.4 $ 21, Public Education Total $ 3,052.6 $ 12, Higher Education Total $ $ 3, Social Services Total $ $ 2, Infrastructure Total $ $ 1, Buildings $ - $ - Debt Service $ $ 1, Transportation $ (5.1) $ (12.21) Criminal Justice Total $ $ 1, Economic Development Total $ $ Environment and Natural Resources Total $ $ Elected Officials Total $ 86.6 $ General Government Total $ 28.0 $ $14,000 $12,000 $10,000 $8,000 $6,000 $4,000 $2,000 $0 $11,961 $3,337 $2,072 $1,515 $1,032 $337 $608 $174 $66

23 Revenue Simulator

24 Pitfalls and Problems

25 Reliance on External Forecasts Can Be Problematic Here, La wages & salaries income. History sourced to the BEA, but forecasts provided by Moody s Analytics. The Moody s forecast of LA income (La variables in general) has been persistently more optimistic than has actually prevailed. Thus, the Moody s series has to be adjusted throughout the forecast horizon to something more realistic/likely. A simple extrapolation of underlying trend path or near-term historical performance provides a more likely forecast of this variable for use in sales tax forecasting. {Moody s forcast implies annual growth rates up to 5% before tapering off; Adjusted rate no more than 2.5%} La Wages & Salaries Income Historical and Forecasts qtrly obs, SAAR Moody's Forecast Adjusted Moody's Forecast Q1 1990Q4 1991Q3 1992Q2 1993Q1 1993Q4 1994Q3 1995Q2 1996Q1 1996Q4 1997Q3 1998Q2 1999Q1 1999Q4 2000Q3 2001Q2 2002Q1 2002Q4 2003Q3 2004Q2 2005Q1 2005Q4 2006Q3 2007Q2 2008Q1 2008Q4 2009Q3 2010Q2 2011Q1 2011Q4 2012Q3 2013Q2 2014Q1 2014Q4 2015Q3 2016Q2 2017Q1 2017Q4 2018Q3 2019Q2 2020Q1 2020Q4 La Legislative Fiscal Office

26 Reliance on External Forecasts Can Be Problematic Same issue here, but with regard to oil prices. Moody s doesn t have to live on any of its forecasts, but we have to live on ours. External forecasts of oil price seem to be more aggressive than what we are comfortable budgeting. So we adjust them down (an average of Moody s, EIA, our Natural Resources Dept. low price scenario, and sometimes NYMEX futures) Moody's Adjusted Moody's Oil Price :1 91:1 92:1 93:1 94:1 95:1 96:1 97:1 98:1 99:1 00:1 01:1 02:1 03:1 04:1 05:1 06:1 07:1 08:1 09:1 10:1 11:1 12:1 13:1 14:1 15:1 16:1 17:1 18:1 19:1 20:1 21:1 La Legislative Fiscal Office

27 Even When Good Data Is Available, It Can Be Used Badly The 2015/16 Executive Budget proposed to balance the total proposed spending with existing baseline revenue + the recommended additional revenue below from curtailing usage of certain tax credits. The additional revenue figures were taken from the Revenue Dept. annual publication titled Tax Exemption Budget, reporting actual credit usage. from 2014 teb Amount Offsetting Tax Amount Refunded La Legislative Fiscal Office Inventory Tax Credit Ad Valorem Credit for Certain Natural Gas Ad Valorem Credit for Offshore Vessels Telephone Co Property Tax Credit 452,676,421 4,534,210 56,406,978 22,643,842 75,961, ,179 13,356,040 17,381, ,715,230 3,983,031 43,050,938 5,261,969 A bill addressing these 4 credits expected a fiscal note reflecting $429.1 million of additional revenue for FY15/16 (82% of the $525.9 million total recommendation). This estimate, used in the budget, came directly from Revenue Dept. data concerning the utilization of these credits. To the left, the latest year available of a multi year table of credit use for these 4 credits and all the others mentioned above (from the Revenue Dept.). = $429.1 million

28 Even When Good Data Is Available, It Can Be Used Badly The budget expected $429.1 million for FY15/16 from these 4 credits. This is the fiscal note they got. A little short of that. What went wrong? The administration looked at the figures provided by the Dept in the Tax Exemption Budget. That document reports exemption losses (credit losses in this case) by state fiscal year. However, the bill reducing the credits was made effective (appropriately so) on the basis of tax/calendar year, Applicable to tax years beginning on and after January 1, No one in the administration seemed to realize this until the fiscal note was generated. That note explained it, as nicely as possible, as follows: So what did we do? In addition to many other changes that made the bill almost unrecognizable from the Original version, the effectiveness / applicability language was changed to: Applicable to all claims for these credits on any return filed on or after July 1, 2015, regardless of the tax year to which the return relates. La Legislative Fiscal Office

29 Even When Good Data Is Available, Implementation Can Make Estimates Useless Table below is a sample of near-term history of tax credit usage going into the 2015 legislative session. For many credits history actually went back many years prior. Very good data on actual credit use, annual growth & variation. from 2014 teb Amount Offsetting Tax Amount Refunded Citizens Insurance Credit Inventory Tax Credit 41,100,487 33,546,054 7,554,433 Ad Valorem Credit for Certain Natural Gas Ad Valorem Credit for Offshore Vessels Telephone Co Property Tax Credit 452,676,421 4,534,210 56,406,978 22,643,842 75,961, ,179 13,356,040 17,381, ,715,230 3,983,031 43,050,938 5,261,969 Milk Producers Technology Commercialization Historic Residential Rehab Angel Investor Broadway South 1,555, , ,657 1,568,555 8,754, ,401 27, , , ,497 1,443, , , ,657 8,569,107 School Readiness ChildCare Provider 4,662, ,260 4,406,296 Sugarcane Trailer Conversion 2,744,431 23,053 2,721,378 Research & Development 24,380, ,852 23,585,961 from 2013 teb Amount Offsetting Tax Amount Refunded 40,697,004 33,012,766 7,684, ,306,842 4,022,555 41,457,576 24,097,188 77,639,885 1,216,471 5,263,692 15,204, ,666,957 2,806,084 36,193,884 8,892,202 1,810, , ,818 1,819,274 4,950, ,934 2, , , ,943 1,702, , , ,729 4,748,073 5,506, ,417 5,228,403 5,733, ,868 5,365,825 19,703,087 1,444,225 18,258,862 from 2012 teb Amount Offsetting Tax Amount Refunded 42,078,539 33,813,798 8,264, ,004,799 3,276,447 43,768,033 24,425,749 65,086, ,684 8,394,741 17,817, ,917,977 2,409,763 35,373,292 6,607,857 1,737, , ,476 3,386,710 5,261, ,346 20, ,380 1,276,181 76,631 1,612,154 84, ,096 2,110,529 5,185,352 5,617, ,312 5,355,797 1,746, ,202 1,598,940 10,795, ,834 9,824,404 from 2011 teb Amount Offsetting Tax Amount Refunded 43,527,560 35,421,468 8,106, ,990,196 5,407,465 37,454,767 69,304,257 83,704,067 1,108,439 10,254,776 26,628, ,286,129 4,299,026 27,199,991 42,676,241 1,778,750 91, ,435 3,493, , ,587 8, ,909 1,371,238 29,583 1,645,163 83, ,526 2,122, ,080 4,685, ,578 4,463, ,335 91, ,603 12,449, ,029 11,769,230 from 2010 teb Amount Offsetting Tax Amount Refunded 46,284,427 37,818,083 8,466, ,710,932 3,825,605 29,795,877 23,306,138 57,553,014 1,129,622 10,430,960 18,386, ,157,918 2,695,983 19,364,917 4,919,662 1,885, , ,553 2,664,283 30, ,048 1, ,483 1,037,563 21,888 1,736, ,481 68,070 1,626,720 8,136 2,818, ,711 2,674,689 This data and other data on actual usage of credits, deductions, programs was used to pass legislation with estimated effects below. Table 1 Summary of Major Revenue Bills, 2015 Session (in millions) 197,642 32, ,986 2,262,509 62,683 2,199,826 FY16 Duration Applicable Recoupment Act 125 / HB 629 Income & Franchise Tax Credits Cut 28% $ years All Returns From July 1 3 years Act 123 / HB 624 Corporate Income Tax Exclusions and Deductions Cut 28% $ years All Returns From July 1 3 years Act 133 / HB 805 Five Year Carry-forward of 25% of Inventory Credit $129.0 Permanent All Returns From July 1 None HCR 8 Suspend Business Utilities Exemption to 1% of Sales Tax $ / 27/ 16 Transactions From July 1 None Act 94 / HB 119 Increase Cigarette Tax by 50 / pack plus vapor products $106.4 Permanent Transactions From July 1 None Act 109 / HB 402 Equalize Credit for Taxes Paid to Other States $ years All Returns From July 1 3 years Act 103 / HB 218 Eliminate Net Operating Loss Carry-Backs $29.0 Permanent All Returns From July 1 None Act 131 / HB 779 Cap Solar Tax Credit Program $19.0 Permanent All Claims From Jan 1 None Act 126 / HB 635 Enterprise Zone Restrictions $5.0 Permanent All Claims From July 1 None Act 134 / HB 829 Modify / Cap Film Tax Credit Program $ years All Claims From July 1 None Act 110 / HB 445 Certificates Of Title Tax Increase $59.5 Permanent Transactions From July 1 None Act 147 / SB 271 Reduce Motor Fuels Tax Remitance Discounts $6.0 Permanent Transactions From July 1 None Act 109 / SB 93 Prohibits Education Credit If Tuition Deduction Taken $2.3 Permanent From Tax Year 2015 None Interaction Between Act 123 and Act 103 ($8.0) Total Additional Revenue Generated $719.9 plus $46.8 million in new fees = $766.6M new money raised Ultimately, we couldn t tell if any of this revenue actually materialized in FY16. Much of it ($404M) was to show up in corporate tax collections. Beginning-of-year forecast = $790M vs end-of-year actual collections = $248M. Look at implementation provisions: (a) 3-year temporary duration (wait us out and receive 100 on the dollar), (b) applicable to returns received after 7/1/2015 regardless of tax year of the filing (file quickly and receive 100 on the dollar), (c) 3 years to recoup extra tax payments if return filed after 7/1/15 was a final return of an earlier extension (this made the accountants happy). La Legislative Fiscal Office

30 Lack of data can lead to data collection effort (FY 2016/17) Sales tax rate increase and base expansion to tax previously exempt transactions estimated based on current collections from the traditional base, and on data from the Tax Exemption Budget for exempt transactions. Only a handful of sales tax exempt transactions are reported in the Exemption Budget b/c there are so many of them. Some 90+ exempt transactions reported in a catch-all line, All Other Exempt Sales. A very large number implying a very large amount of tax receipt should those transactions be taxed. Fiscal note on the bill assumed only 10% of the implied tax amount from taxation of those transactions. Very controversial assumption. Total sales tax collections for the year came in $30+M less than forecast, inclusive of the fiscal note. Revenue Dept implemented a tax form that attempts to capture all the catch-all transactions individually. La Legislative Fiscal Office