UKMOD training course Mike Brewer Paola De Agostini

UKMOD training course Mike Brewer, Paola De Agostini, Iva Tasseva and Kakia Chatsiou ISER, University of Essex 25 th October 2019 Course based on UKMOD_A 1. 0+

2 Welcome! • Day 3: o o Part 1: New features in UKMOD A 1. 0+ Part 2: Working with UKMOD – advance uses Updating 2020/21 system following the Budget Ø Policy Effect Tool Ø Extra slides: Other Tools available with the EUROMOD UI: MTR add-on Ø

3 Course structure Mix of lecture and hands-on exercises Extra exercises offered to do after the course Please ask questions, and definitely interrupt if terminology is unclear Tell us what you would like to do with the model

The Nuffield-funded project’s objectives 1. Improve EUROMOD-UK’s relevance for UK policy debates (by adding new functionality) -> UKMOD 2. Widen access through e. g. demonstration projects, better documentation, a support group, simpler interface 3. As a case study, show UKMOD can be used to design a Basic Income scheme 4. Suggest an agenda for “where next? ” that might include webhosting of the model, and its potential use in education

7 What is today about? … • getting to know the new features included into UKMOD: o o • o Data Policies Systems Interface elements needed for interpretation of results a general understanding of how to use the software for your own analyses

An Introduction to the new features of UKMOD A 1. 0+ Day 3 Part 1

10 What’s new in A 1. 0+? • Countries/Nations o • Tax-benefit rules (systems) o o • UK, Scotland Wales 2019/20 2020/21 Input datasets Ø Ø o Countries bundle Pooled data: Ø Ø • UK_2016_a 2 from FRS 2016/17 UK_2017_a 1 from FRS 2017/18 UK_2016_b 3: pools FRS 2014/15, 2015/16 and 2016/17 UK_2017_b 1: pools FRS 2015/16, 2016/17 and 2017/18 Tax-benefit policies rules o o Universal Credit Scottish Carer’s Allowance Supplement Scottish Best Start Grant Devolution of income tax extended to Wales • Software version EUROMOD Users Interface version 3. 1. 13 • Output analysis tool: o Renew Statistics Presenter Tool EUROMOD software

11 Part 1 - Outline 1. New policies: Universal Credit 2. New countries/nations files 3. Projected 2020/21 system

13 1 - Learning about Universal Credit: outline • • About Universal Credit Position in UKMOD policy spine (bsauc_uk) Universal Credit: order and logic of calculation o Eligibility o Benefit amount Assumptions & Options: o Non Take-up correction o Transition from legacy benefits to universal credit o Extension tools: LBA, UCA and BTA

Universal Credit (UC) • Means-tested benefit for low income working-age families • Replacing: WTC, CTC, IS, ib-JSA, ib-ESA, HB • Working-age families (identified as tu_bu_uk) where at least one adult’s age under State Pension Age (SPA) • => couples (joint claim) where one over SPA and one under SPA are considered under UC rules instead of PC • Gradual introduction started in 2013 (pilot) and will carry on until 2022/23

UC: position in the policy spine (bsauc_uk) After: WTC/CTC (bwkmt_bfamt_uk), IS/JSA (bsa_uk), ESA (bsadi_uk) and HB (bho_uk); Before: CTR (bmu_uk)

16 UC: order and logic of calculation • • Eligibility rules Benefit amount calculation

UC: Who can claim – eligibility conditions (1/3) • Basic conditions Ø Age rule: individuals 18<=dag<$SPAfem (qualifying age for PC) Ø Education rule: not in education Ø Have accepted a “claimant commitment” (conditionality rules) • Financial conditions Ø Low income Ø Savings and other capital not more than £ 16, 000

UC: Who can claim – eligibility conditions (2/3) Exceptions: • To age rules: Ø 16/17 year olds in special cases such as parents, disabled people or young people estranged from their parents Ø Over $SPAfem living with partner in working age group • To education rule: Ø parents, disabled students or young people on non-advanced courses without parental support • For couples (joint claim): Ø In general both must satisfy basic conditions Ø BUT exceptions Ø one over $SPAfem or Ø one in education

UC: Who can claim – eligibility conditions (3/3) UKMOD implements eligibility conditions in various steps: 1) Identifies parents (i_bsauc_elig 1=1) 2) Working-age individuals not in education OR in education but receiving some disability benefits (i_bsauc_elig 2=1) 3) Exceptions: 16/17 y. o. with disability or parents; over $SPAfem with a young partner (i_bsauc_elig 3=1) 4) Combines above with financial condition (savings and capital) (i_bsauc_elig=1)

UC: benefit amount Four steps calculations: 1. Calculate maximum amount (basic amount + premiums) 2. Work out earnings and check how much can be ignored 3. Work out other income and how much can be ignored 4. Final UC entitlement: reduce maximum amount by relevant means

1) Calculate maximum amount + Standard allowance (different amounts for single and for couples) + child element for each child, with an increase for disabled and severely disabled children (two-child limit applies here from April 2017) + limited capability for work element or work-related activity element (abolished for new claims after April 2017) + carer element + housing costs element (from April 2018 not covering mortgage) + childcare cost element

2) Earnings and how much can be ignored • Work allowance (ignored earned income): Eligibility: Ø People responsible for a child OR Ø People with limited capability to work Amount: Ø Depends on whether UC includes housing costs element • Earned income exceeding work allowance reduces UC max amount by 63 p per extra £ 1 (taper 63% from 2017) • Only one allowance applies per family • UKMOD relevant income list is il_uc_means_earned

3) Other (unearned) income • Ignored/disregarded income: Ø e. g. child benefit and personal independence payment • Counting in full: Ø Other income - e. g. occupational and personal pensions • Capital over £ 6, 000 is treated as producing an income of £ 4. 35 a month for every £ 250, or part, above the limit (tariff income). • Unearned income reduces UC pound-by-pound • UKMOD relevant income list is il_uc_means_unearned

4) Final UC amount UC max amount - unearned income - 63% of earned income minus work allowance ____________ Final UC amount • UKMOD output variable bsauc_s

25 • Relevant UC parameters in UKMOD Defined by policy Const. Def_uk

Things to keep in mind: • HB still working for non-working-age people => set to zero only for working-age households • Interactions of UC with the rest of T-B system: • Enter means for CTR (bmu_uk) • Capped instead of HB (bcap_uk)

27 Exercise 1 • You will: o Implement a reform to Universal Credit(bsauc_uk), based on UK 2019 system o Identify the part implementing the two-child limit within UC Ø Ø Ø Compare with UK 2018 system to see how this policy has been introduced Make a copy of UK 2019 system and call it e. g. UK_2019_ex 1 Modify UK_2019_ex 1 – Remove the two-child limit o Run reform system and analyse impact on household incomes of the reform using the Statistics Presenter

28 Summary: Exercise 1 • You learned o o how UC is implemented in UKMOD how to modify UC by removing the two-child limit policy introduced for new claimants

29 Questions

30 UKMOD Universal Credit: assumptions and options • Baseline assumptions o o • Transition from legacy benefit to UC Non take-up corrections for means-tested benefits (partial takeup) Options o o o Full legacy benefits Full Universal Credit No take-up corrections (full take-up)

31 UC: transition from legacy benefits • Baseline assumes transition from legacy benefits to UC o Based on random selection o Follows DWP and OBR statistics and forecasts • Baseline: Const. Def_uk defines $UCtransition • Optional: 2 extensions (access from running window) allow users to chose: o o Legacy Benefits Adjustment (LBA): assumes full legacy benefits (no UC) Universal Credit Adjustment (UCA): assumes full universal benefit (no legacy)

32 Select extensions UKMOD: how to run extensions (switches) Select OPTIONS When ON: • LBA sets $UCtransition = 0 => no transitions to UC • UCA sets $UCtransition = 1 => full roll out of UC

33 UC: non-take-up correction • UKMOD baseline allows for Non-Take-up corrections: o o Extension Benefits Take-up Adjustment (BTA) set to ON by default It distinguishes 3 groups: Ø Ø Ø • Need information about legacy benefits entitlement and takeup behaviour: o Relevant income lists: Ø Ø • Takers of (at least one) legacy benefit replaced by UC: if entitled to UC, assume they will still take-up No takers of (any) legacy benefit replaced by UC: if entitled to UC, assume they will not take-up New entitlements: assume IS take-up rate Il_legacy_elig: sum of legacy benefits before take-up corrections apply Il_legacy_tu: sum of legacy benefits after take-up corrections apply UKMOD NOTE: some extensions can be combined (i. e. BTA = OFF & UCA = ON produces simulations for a full UC & full take-up scenario).

34 Exercise 2 • You will learn: o How to use extensions LBA and UCA Ø BTA Ø Combining them Ø

35 Summary: Exercise 2 • You learned how to produce results for the following scenarios: o o Baseline: partial transition from legacy benefits to UC LBA on: no transition to UC UCA on: full UC roll-out BTA off: full take-up

36 Questions

37 2. New countries/nations files • Separate nations xml: SCT & WLS o • Using pooled datasets uk_2016_b 3 and uk_2017_b 1 o o • … but also extensions: SCT_? ? and WLS_? ? (see next slide) Adjusted Uprating function Adjusted sample weights (dwt) Specific nations’ policies: o o o Carer’s allowance supplement Best Start grant Devolution of Income tax

38 …but also country specific extensions

39 Summary Countries/Nations Extensions • UK default result is UK_yyyy_std. txt o o • SCT default (SCT= ON) result is SC_yyyy_std. txt o o • If SCT OFF => SC_yyyy_SCToff. txt Ø This is the same as UK_yyyy_std. txt If SCT OFF & WLS ON => SC_yyy_SCToff. WLSon. txt Ø This is the same as WL_yyyy_std. txt WLS default (WLS = ON) result is WL_yyyy_std. txt o o • If SCT ON => UK_yyyy_SCTon. txt Ø This is the same as SC_yyyy_std. txt If WLS ON => UK_yyyy_WLSon. txt Ø This is the same as WL_yyyy_std. txt If WLS OFF => WL_yyyy_WLSoff. txt Ø This is the same as UK_yyyy_std. txt If WLS OFF & SCT ON => WL_yyyy_WLSoff. SCTon. txt Ø This is the same as SC_yyyy_std. txt UKMOD NOTE: some extensions can NOT be combined (i. e. SCT = ON & WLS = ON produces => all data in memory dropped).

40 New pooled datasets • New pooled datasets uk_2016_b 3 and uk_2017_b 1 available with this release Ø Ø • UK_2016_b 3: pools FRS 2014/15, 2015/16 and 2016/17 UK_2017_b 1: pools FRS 2015/16, 2016/17 and 2017/18 Data adjustments (already done: no need to do anything) o Uprating function: use of variable dpd “data period” o Sample weights (dwt): adjusted to make the data sample representative of the population

41 New “regional” policies • • • Devolution of Income tax to Scotland Wales (tin_uk) Scottish carer’s allowance supplement (bcas_uk) Scottish Best Start grant (bmascmt_uk) • See detailed description in Country Report

42 Exercise 3 • You will learn: o How to reform income tax for Wales o Open the WL model Analyse how the Welsh population is selected in WLS_wl Analyse the structure of the personal income tax in the WLS focusing on the personal allowance (tinta_wl) and the tax schedule (tin_uk) Reform the 2019 system and introduce a separate tax schedule for Wales which adopts the same income bands as Scotland o o o Use UK_2017_b 1 data Run reform system and analyse impact on household incomes of the reform in Wales using the Statistics Presenter

43 Summary: Exercise 3 • You learned o o o how to produce results for a region in the UK how to modify the personal income tax calculation for a region of the UK how to use pooled data

44 Questions

45 3. Projected 2020/21 system • UKMOD A 1. 0+ includes 2020/21 based on: o Policy reforms announced up to March 2019 Spring Financial Statement Ø o OBR forecasts from Fiscal and Economic outlook from March 2019 (i. e. CPI, AEI, triple lock) Ø Ø • Structural policy changes (i. e. two-child limit for new claimants) Data uprating (function Uprate) Parameters projections (see policy Const. Def_uk) UKMOD note: 2020/21 system needs to be updated following the next Budget!

46 Updating the projected 2020/21 system • Include new policy reforms o o • Structural (i. e. introduction of a new benefit) Parametrical (i. e. change of existing “projected” benefit amount) Update uprating indexes o To take into account new OBR projections

47 UKMOD: SYSTEM FUNCTIONS Uprate

48 System Functions: Uprate • Indices to uprate monetary variables to price level of policy year name of variable to be uprated value of uprating factor

49 Uprating indices (1) • • • Datasets -> used with more than 1 system o May be needed for various exercises (ex: Policy Effects) o Fewer datasets than systems Discrepancy between the year the dataset refers to and the policy year Use uprating indices to index the incomes in the dataset to the policy year (i. e. the year you want to simulate) o Ex: index the income data referring to 2011 to the 2014 level Uprating indices- based on raw series EM calculates implicit updating factors

50 Uprating indices (2) policy years the index name the indexreference (handle) ($f_) source of information

51 Uprating indices (3) • You can see the implicit updating factors applied • Select the dataset you want to view • Example: • Dataset UK_2017_a 1 income referring to 2017 • CPI increased between 2017 & 2020 by approx. 6%

52 Using Uprate • Uprating Indices used by the function Uprate Name of income variable to be uprated Reference of index to be used for uprating Dataset(s) to which the function applies

53 Using Uprate • Aggregate variables • Using different values for different groups

54 Exercise 4 • You will learn: o How to include CPI projection changes using the uprating indices table

55 Summary: Exercise 4 • You learned o how to update projected 2020/21 following an announce change in CPI

56 Questions

57 Advance uses of UKMOD • • Policy Effect Tool (PET) Marginal Effective Tax Rate (MTR) add-on (extra slides)

58 In this lecture, you will learn: • How to calculate the effect on the income distribution of policy changes, using the Policy Effects Tool o Example question: what would be the average income, if, instead of 2019 policies, 2020 policies were in place? o Abstract from (i. e. ignore) changes due to changes in pretax earnings or population characteristics to focus only on policy changes o Have to decide “what counts as a policy change? ” Ø Ø Any nominal change in cash values? Any change in cash values different from price inflation?

59 Policy Effects Tool (PET) (1) • PET calculates the effect of policy changes as (b-a): a. b. Household net incomes under year 1 tax-benefit policies Household net incomes under year 2 tax-benefit policies Ø Ø • No change to tax-benefit parameters: nominal difference Or deflating tax-benefit parameters back to year 1 by CPI: real difference Example question: what would be the average income, if, instead of 2019 policies, 2020 policies were in place? Data on market incomes & population characteristics 2019 data (FRS 2017/18 data uprated to 2019) Tax-benefit policies Household net incomes 2019 policies (a) 2019 baseline 2020 policies: - no change or - deflated back to 2019 prices (b) 2020 counterfactual Effect of policy changes (b) - (a)

Policy Effects Tool (PET) (2) • Nominal vs real changes o Example A: personal allowance freeze Ø Ø o Example B: Benefit increase – smaller than inflation Ø Ø • Nominal changes: person A has gross income of £ 20, 000. As the Personal Tax Allowance is frozen bw 2019 and 2020, A’s net income does not change. Real changes: CPI bw 2019 and 2020 is 1. 9%, so PTA has gone down in real terms. In real terms, A pays more taxes under 2020 policies than under 2019 policies, so A’s net income is smaller. Nominal changes: the Winter Fuel Allowance is increased by 1% bw 2019 and 2020. Person B receives WFA. Thus, B’s net income goes up. Real changes: CPI bw 2019 and 2020 is 1. 9%, so higher than the 1% increase in WFA. In real terms, WFA has gone down and B’s net income is smaller. Although we do not simulate all benefits (e. g. pensions) and taxes (e. g. council tax), we use uprating factors to capture statutory indexation and ad-hoc changes o e. g. the Basic State Pension is uprated by triple-lock indexation bw 2019 and 2020 Ø Ø Nominal change: increase in the Basic State Pension Real change: depends on the triple-lock indexation value vs inflation 60

61 Policy Effects Tool (PET) (3) • Effect of policy changes o expressed as % of mean household net income in year 1 o by decile group and for the population as a whole Ø Ø o individuals ranked by equivalised household net income in year 1 modified OECD scale (1: 0. 5: 0. 3) decomposed by tax-benefit policy • See Annex 2 of the UKMOD Country Report

62 Exercise 5 • You will learn: o o How to use the Policy Effects Tool (PET) How to interpret the results • We will do the exercise together! o Each step is described in the following slides if you wanted to go back to them later

63 Opening PET

Options in PET select policy years 1 and 2, always two consecutive years • • input data on gross market incomes and population characteristics if input data and policy year differ, market incomes brought to policy year 1 through uprating • • CPI (= $HICP in uprating factors table) real changes to taxes and benefits 1 nominal changes to taxes and benefits 64

65 Results in PET (1) real changes to taxes and benefits, CPI value displayed change in mean net income decomposed by income type change in income by decile group and for the population as a whole change in mean net income

Results in PET (2) no change to original/gross market income by design 66

67 Results in PET (3) • Effect of the 2019 -2020 policy changes on net income: o o reduction of 0. 2% on average progressive: gains or no change at the bottom and losses in the middle and top of the distribution • Decomposition by taxes and benefits o o income gains due to public pensions and meanstested benefits, for the bottom deciles income losses due to increases in direct taxes

Results in PET (4) • Changes to income taxes (ils_tax) o income loses across the entire income distribution Ø o • income gains due to lower NI contributions Changes to public pensions (ils_pen) o o • progressive as income losses increase with decile group Changes to employee National Insurance contributions o • Personal Tax Allowance and high tax threshold frozen in nominal terms falling in real terms income gains across the entire distribution triple-lock indexation (captured by the uprating) higher than growth in prices Changes to means-tested benefits (ils_benmt) o increase in net income for deciles 1 and 3 due to benefit increases Ø Ø o higher share of claimants transferred to UC gains to higher take-up of UC compared to the legacy benefit system small income losses for almost all other deciles 68

Benefit take-up • PET results capture changes to benefit take-up rates o o • changes to the take-up of means-tested benefits and tax credits changes to the share of claimants who transfer from the legacy system to Universal Credit To change take-up assumptions for PET: o PET results assuming full take-up of means-tested benefits and tax credits Ø o PET results assuming no change to the share of claimants of UC Ø o Country tools set switches BTA change from “on” to “off” for relevant data/system combinations press ok save Open country model policy Const. Def_uk function Def. Const (“UC roll-out: DWP transfer share estimation”) change constant $UCtransition save Remember to reverse changes to the model when you finish with PET 69

70 Summary: Exercise 5 • You learned: o how to use PET to analyse effect on the income distribution of policy changes o how to interpret PET results

71 Extra slides

72 METR – marginal effective tax rate

Incentives to work more/earn more •

How to calculate METR in UKMOD (1/2) • MTR Add-on (part of UKMOD A 1. 0+ version) Click “View Add-ons” and Select MTR

How to calculate METR in UKMOD (2/2) • MTR Add-on (part of UKMOD A 1. 0+ version) Select the system/data for which you want to run MTR

METR in UKMOD: assumptions • • Simulates increase of earnings by 3% (can be changed in the Add-on) o 3% roughly corresponds to an extra working hour per week Focuses on individuals (18 -65) currently in work with positive earnings Assumes full compliance (benefit full take-up) «intended effect» of the tax-benefit system on labour market incentives Static effect: Assumes that behaviour of other household members does not change when a person increases his/her working hours/earnings

77 METR in UKMOD • Output o Micro-data files cc_yyyy_mtr. txt • Further information on how METR are calculated in UKMOD and how to use the add-on can be found in the EUROMOD User Interface help • UKMOD note: this data must be analysed with some statistical package outside EUROMOD User Interface