AboutSubmit my research

Methodology, Taxes and benefits

Research articles · Aug 31, 2016

Improving the validity of microsimulation results: Lessons from Slovakia

Zuzana Siebertova, Norbert Svarda, Jana Valachyova

Council for Budget Responsibility I. Karvasa 1, Slovakia; Faculty of Mathematics, Physics and Informatics Comenius University, Slovakia

Volume 9(2) Summer 2016

Cite as: Z. Siebertova, N. Svarda, J. Valachyova; 2016; Improving the validity of microsimulation results: Lessons from Slovakia; International Journal of Microsimulation; 9(2); 77–105. doi: 10.34196/ijm.00136

Abstract

This paper summarizes the lessons learned in the process of building a microsimulation tool tailored to country-specific conditions and involving a maximum degree of user control. The objective to construct a model useful in the process of budgeting and fiscal forecasting has been achieved by paying attention to policy simulation details as well as to the representativeness of the underlying micro-dataset. The validity of simulated results improved significantly after the input database sample has been reweighted in such a way that the new weights replicate, among other factors, the earned income distribution and selected age cohorts directly. Innovative approaches in bringing the model closer to legislation as well as data highlight the benefits of having more user control compared with standardized microsimulation tools.

This article contains 1 figure and 14 tables.

Figures

Figure A1. Income distribution SK-SILC and social security agency (SSA).

Tables

Table 1. Sources of external controls for calibration.
Calibration category Source of the official statistics
Age cohorts (0, 1–3, 4–16, 17–25, 26–45, 46-retirement age, over retirement age). Categories over 16 years considered separately for gender (male, female) Statistical Office of the Slovak Republic (Demographic Balances of Population)
Labour market status (employees, unemployed, self-employed) Statistical Office of the Slovak Republic (Labour Force Survey)
Household size (members 1, 2, 3, 4, 5+) Statistical Office of the Slovak Republic (Population and Housing Census 2011)
Gross income from employment distribution (deciles 1–2, 3–5, 6–8 and 9–10) Database of individual records from Social Security Agency (SSA)
Table 2. Descriptive statistics of grossing-up weights in SK-SILC samples.
Policy year 2011 2011 2012 2012 2013 2014
Underlying SILC dataset 2011 2011 2012 2012 2012 2012
Grossing-up sample weight Original Calibrated Original Calibrated Original Calibrated
Mean 349.5 350.0 350.4 350.0 351.2 351.8
Std. Dev. 126.0 468.7 132.0 482.5 521.3 581.1
Minimum 108.7 10.9 119.9 12.0 12.0 12.0
Maximum 1,226.1 5,505.1 1,083.9 4,481.0 5,974.1 8,961.1
Dataset characteristics
Individuals 15,440 15,440 15,426 15,426 15,426 15,426
Households 5,291 5,291 5,402 5,402 5,402 5,402
Projected population 5,395,519 5,396,355 5,404,664 5,398,917 5,417,340 5,427,220
Projected households 1,911,664 1,909,248 1,852,059 1,852,027 1,852,059 1,853,409

Source: Author’s calculations using SK-SILC.

Table 3. Validation of weighting schemes: representation of gender and age cohorts.
Original weights Calibrated weights
2011 2012 2011 2012
Female 1.00 1.00 1.00 1.00
Age cohort
0 0.55 0.78 1.00 1.04
0–3 0.63 0.84 1.00 1.01
0–16 0.82 0.96 1.00 1.00
0–26 0.98 1.04 1.00 1.00
Prime age 1.03 1.01 1.00 1.00
Retirement age 1.08 1.04 1.02 1.01

Source: Authors' calculations using SK-SILC and Slovak Statistical Office

Note: Ratios display number of individuals in SK-SILC dataset (weighted) to population by gender and in the respective age cohort.

Prime age: 15–64 years. Retirement age: males 62+, females 58+ in 2011 and 59+ in 2012.

Table 4. Validation of weighting schemes: representation of economic activity.
Original weights Calibrated weights
2011 2012 2011 2012
Employed 1.01 0.97 0.96 0.96
Employee 1.20 1.15 1.08 1.08
Self-employed 1.01 0.96 1.02 1.01
Unemployed 0.97 0.94 0.99 0.99
Economic active pop. 1.01 0.97 0.97 0.97
Economic inactive pop. 0.99 1.03 1.04 1.04
Population total 1.00 1.00 1.00 1.00

Source: Authors’ calculations using SK-SILC and LFS.

Note: Ratios display number of individuals in SK-SILC dataset (weighted) to LFS in the respective category.

Table 5. Validation of weighting schemes: representation of income.
Amounts (ratios) Individuals (ratios)
Weights Original Calibrated Original Calibrated
Income from 2011 2012 2011 2012 2011 2012 2011 2012
Employment 1.16 1.02 0.97 0.95 1.08 1.05 0.93 0.95
Agreements 0.43 0.34 0.43 0.34 0.36 0.38 0.33 0.36
Self-employment * 2.84 2.73 2.98 2.78 1.34 1.25 1.36 1.31
Employment and agreements 1.13 0.99 0.94 0.91 0.97 0.94 0.97 0.94

Source: Authors' calculations using SK-SILC and SSA.

Validation of income for self-employed is only indicative. SK-SILC reports for self-employed the value of profit/loss in the current year, while the SSA database reports the assessment base which is based on the value of return in the year t-2 (inconsistency both in variable and time).

Note: Amounts (ratios) display aggregate amount of income of individuals in SK-SILC dataset (weighted) to aggregate income computed by using records from SSA. Individuals (ratios) display aggregate number of individuals in SK-SILC dataset (weighted) to aggregate computed by using records from SSA.

Table 6. Non-simulated benefits and pensions.
Amounts (ratios) Recipients (ratios)
Weights Original Calibrated Original Calibrated
Benefits 2011 2012 2011 2012 2011 2012 2011 2012
Maternity 0.63 0.60 1.22 1.27 0.57 0.61 1.01 1.28
Sickness 0.31 0.34 0.50 0.59 n.a. n.a. n.a. n.a.
Pensions
Old age 1.16 1.09 1.09 1.08 1.10 1.04 1.04 1.03
Disability 0.87 0.88 1.03 1.10 0.84 0.86 1.03 1.10
Widow/er 0.98 1.05 0.97 1.06 1.06 0.99 1.01 0.97
orphans 0.83 0.82 0.93 0.57 0.69 0.68 0.79 0.52

Source: Authors' calculations using SK-SILC and SSA.

Table 7. Simulated benefits: Ratios of aggregate amounts.
Model EUROMOD SIMTASK
Weights Original Calibrated Original Calibrated
2011
Unemp.benefit 0.54 0.46 0.55 0.47
Parental allowance 0.78 1.18 0.64 0.99
Child benefit 0.87 1.00 1.08 1.16
Child birth grant 0.61 1.10 0.61 1.10
Material needs benefit 1.24 1.82 0.83 1.19
2012
Unemp.benefit 0.38 0.37 0.40 0.38
Parental allowance 1.09 1.07 0.87 0.96
Child benefit 0.97 0.98 1.17 1.14
Child birth grant 0.80 1.08 0.81 1.08
Material needs benefit 1.46 1.92 1.00 1.27

Source: Authors' calculations using SIMTASK and EUROMOD, official statistics SSA (unemployment benefit), other benefits COLSAF (Central Office of Labour, Social Affairs and Family).

Note: Numbers display ratios of aggregate number of payments to individuals computed by EUROMOD and SIMTASK to aggregate amounts referred by official statistics.

Table 8. Personal income tax and social insurance contributions: Aggregate amounts.
Model EUROMOD SIMTASK
Weights Original Calibrated Original Calibrated
2011
Personal income tax 1.17 1.05 1.14 1.03
Social insurance contrib. (SIC)
SIC: Employer 1.24 1.03 1.23 1.02
SIC: Employee 1.25 1.04 1.25 1.04
SIC: Self-employed 1.72 1.77 1.44 1.50
Health Insurance contrib. (HIC)
 HIC: economic 1.19 1.01 1.18 1.00
active pop
 HIC: economic 1.02 1.09 0.94 1.02
inactive pop.
2012
Personal income tax 0.94 0.97 0.91 0.95
Social insurance contrib. (SIC)
SIC: Employer 1.09 1.01 1.09 1.01
SIC: Employee 1.11 1.02 1.11 1.02
SIC: Self-employed 1.63 1.67 1.38 1.41
Health Insurance contrib. (HIC)
 HIC: economic 1.07 1.01 1.06 0.99
active pop.
 HIC: economic 1.03 1.06 0.96 0.99
inactive pop.

Source: Authors' calculations using SIMTASK and EUROMOD, official statistics Ministry of Finance (PIT and HIC), SSA (SIC).

Note: Numbers display ratios of aggregate number of recipients computed by EUROMOD and SIMTASK to aggregate number of recipients referred by official statistics.

Table 9. Simulated benefits (calibrated weights).
Aggregate amounts (I)
2013 2014 2013 2014
Unemp. benefit 0.41 0.48 Child birth grant 1.03 1.21
Parental allowance 0.94 0.92 Material needs benefit 1.32 1.15
Child benefit 1.15 1.15
Recipients (II)
2013 2014 2013 2014
Unemp. benefit 0.48 0.54 Material needs benefit 1.07 1.05
Parental allowance 0.97 0.97 Housing allowance 1.10 1.03
Child benefit 1.08 1.09 Activation allowance 1.59 1.31
Child birth grant 0.98 1.21 Protection allowance 1.07 0.86

Source: Authors' calculations using SIMTASK, official stat. SSA (unemployment benefit), COLSAF (other benefits).

Note: Numbers display ratios of aggregate amount of payments to individuals (I) and number of recipients (II) computed by SIMTASK to official statistics.

Table 10. Personal income tax, social and health insurance contributions (calibrated weights).
2013 2014 2013 2014
Personal income tax 1.00 0.99 SIC employer 0.99 1.00
HIC: economic active pop. 1.04 1.06 SIC employee 1.00 1.00
HIC: economic inactive pop. 0.96 1.02 SIC self-employed 1.42 1.71

Source: Authors' calculations using SIMTASK, official stat. Ministry of Finance (PIT an HIC), SSA (SIC).

Note: Numbers display ratios of aggregate amount of payments computed by SIMTASK to aggregate amounts computed by using official statistics. SIC stands for social insurance contribution and HIC for health insurance contributions.

Table 11. Shares of equalised disposable income by deciles.
EUROSTAT EUROMOD SIMTASK
Original Calibrated Original Calibrated
2011
Decile 1 3.5 3.9 3.7 3.8 3.5
Decile 2 5.7 6.1 5.7 6.0 5.6
Decile 3 6.9 7.5 7.1 7.1 6.8
Decile 4 7.9 8.2 8.1 8.0 7.7
Decile 5 8.6 8.9 8.8 8.9 8.6
Decile 6 9.5 9.8 9.4 9.8 9.5
Decile 7 10.7 10.6 10.5 10.9 10.7
Decile 8 12.1 11.9 11.8 12.1 12.1
Decile 9 14.1 14.0 14.4 14.0 14.6
Decile 10 21.0 19.1 20.5 19.3 20.8
2012
Decile 1 3.6 3.9 3.5 3.9 4.0
Decile 2 5.7 6.2 5.9 6.1 5.4
Decile 3 6.8 7.4 7.1 7.3 7.0
Decile 4 7.7 8.2 8.0 8.2 7.9
Decile 5 8.7 9.0 8.7 9.0 8.7
Decile 6 9.7 9.8 9.6 9.9 9.7
Decile 7 10.9 10.8 10.7 10.9 10.8
Decile 8 12.3 11.9 12.0 12.1 12.3
Decile 9 14.3 13.9 14.3 14.0 14.4
Decile 10 20.3 18.8 20.1 18.7 19.9

Source: EUROSTAT and authors' calculations using SIMTASK and EUROMOD.

Note: The ratio of disposable income in the corresponding decile to the population. Computed for individuals based on household disposable income and equalised by the modified OECD equivalence scale.

Table 12. Income inequality and poverty rates.
EUROSTAT EUROMOD SIMTASK
Weights Original Calibrated Original Calibrated
2011
GINI 25.7 23.3 25.6 23.5 26.0
S80S20 ratio * 3.8 3.3 3.7 3.4 3.9
At risk of poverty rate **
 Total popul. 13.0 12.1 13.4 12.1 13.5
 Females 12.8 12.0 13.5 12.0 13.6
 Males 13.1 12.3 13.3 12.2 13.3
2012
GINI 25.3 22.4 24.4 22.6 24.7
S80S20 ratio * 3.7 3.2 3.6 3.3 3.6
At risk of poverty rate **
 Total popul. 13.2 11.5 12.9 11.7 13.1
 Females 13.2 11.3 12.4 11.5 12.8
 Males 13.3 11.6 13.3 11.9 13.4

Source: EUROSTAT and authors' calculations using SIMTASK and EUROMOD.

The ratio of total income received by the top 20 % of population to that received by the bottom 20%.

Percentage of population below 60% of median equalised income.

Measures are computed for individuals based on disposable household income and equalised by the modified OECD equivalence scale.

Table A1. Summary statistics SK-SILC 2013 (Income reference period 2012).
Mean (Std.Dev) Number of individuals
SK SILC Admin. Data SK SILC Admin. Data
Weights Original Calibrated Original Calibrated
Labour income (monthly)
Gross wage employment 692.97 722.59 805.00 * 1,983,176 1,788,927 1,881,598
(369.69) (530.82) 775.60 **
Income from self-employment 722.89 709.50 382.43 ** 347,397 361,935 277,125
(558.06) (540.66)
Other payments made by employers 17.24 20.79 315,322 278,861
(20.29) (29.89)
Income from agreements 70.16 68.96 143.94 ** 364,458 341,729 955,330
(154.41) (93.76)
Non-labour income (monthly)
Unemployment benefit 271.20 259.96 312.00 58,064 60,399 143,896
(164.21) (148.52)
Maternity benefit 431.87 451.57 443.00 35,203 71,317 24,221
(179.48) (168.34)
Child birth grant 60.46 62.95 153.67 38,020 70,298 56,994
(22.24) (23.17)
Child benefit (incl. additional child benefit) 23.49 23.34 37.78 779,739 738,748 688,344
(12.27) (11.19)
Parental allowance 200.71 200.43 195.87 140,111 164,278 142,274
(78.99) (71.97)
Material needs benefit 132.63 122.99 132.62 102,402 118,566 183,091
(120.49) (103.3)
Nursing allowance 150.05 140.28 142.22 32,411 41,248 58,700
(100.47) (89.04)
Sickness and nursing benefits 78.84 101.03 209.70 107,431 139,511 119,092
(81.78) (96.78)
Disability pension 272.53 261.98 260.90 201,729 257,721 227,801
(112.53) (98.64)
Old-age pension 375.44 374.58 367.00 1,072,056 1,062,519 980,863
(112.07) (108.71)
Early retirement pension 348.33 325.15 374.00 16,765 16,862 24,404
(107.57) (104.03)
Widow’s pension 153.16 156.99 237.00 339,647 346,049 336,877
(97.51) (100.6)
Orphan’s pension 159.09 146.41 126.00 19,479 16,972 26,923
(65.77) (55.18)
Sample size 15,426 15,426

Source: Author’s calculations using SK SILC, official statistics Statistical Office, SSA and COLSAF.

Mean monthly wage reported by the Statistical Office.

Means computed by using data from SSA.

Table A2. Simulated benefits: Ratios of aggregate number of recipients.
Model EUROMOD SIMTASK
Weights Original Calibrated Original Calibrated
2011
Unemp. benefit 0.50 0.44 0.51 0.45
Parental allowance 0.68 1.03 0.67 1.03
Child benefit * 0.88 1.00 1.08 1.15
Child birth grant 0.59 1.07 0.59 1.07
Material needs benefit 1.00 1.36 0.73 1.01
Housing allowance 1.92 2.62 0.79 1.15
Activation allowance 4.68 6.47 1.00 1.50
Protection allowance 1.41 1.87 0.58 0.68
2012
Unemp. benefit 0.38 0.41 0.40 0.43
Parental allowance 0.88 0.87 0.87 0.99
Child benefit * 0.96 0.94 1.15 1.08
Child birth grant 0.75 1.02 0.75 1.02
Material needs benefit 1.07 1.43 0.81 1.04
Housing allowance 2.12 2.78 0.82 1.10
Activation allowance 5.25 6.83 1.25 1.56
Protection allowance 2.06 2.80 0.73 0.92

Source: Authors' calculations using SIMTASK and EUROMOD, official statistics SSA (unemployment benefit).

Official statistics on child benefit recipients is taken as the average of monthly data over the year.

Official statistics on other benefits is the total number of individual recipients (incidence).

Note: Numbers display ratios of aggregate number of recipients computed by EUROMOD and SIMTASK referred by official statistics. COLSAF (other benefits). to aggregate number of recipients.