The OLS regression results for the period 2011 to 2018 are presented in Table 4.1 below. These results show the estimated public/private sector pay differential taking account of when the pension levy is both included in and deducted from gross weekly earnings and when the size of an organisation is both included in and excluded from the model.
Only the estimated public/private sector wage differentials are presented in the tables. More detailed results for other explanatory variables are available in the Detailed OLS Results chapter.
Results for 2011 to 2014 include minor revisions which occur due to the change from the Quarterly National Household Survey (QNHS) to the Labour Force Survey (LFS) and are presented in the table below for information. Commentary in this chapter relates to the results for the period 2015-2018.
The following graphs summarise the results of a series of quantile regression analyses for permanent full-time employees aged 25-59. Regression models both including and excluding size of enterprise were performed and these models were run on earnings after the pension levy was deducted as well as on gross earnings (before deduction of pension levy).
The graphs presented here are based primarily on gross weekly earnings after the pension levy is deducted. Further analysis using different specifications are available on request.
Figure 4.1 shows the premia at various points throughout the earnings distribution (after the deduction of the pension levy) for 2015 to 2018. It is clear that the public sector differential was highest for those at the lower end of the earnings distribution. The pay gap decreased consistently as earnings increased for all four years.
In 2015 the pay differential at the 50th percentile was 0.6% and the percentile at which the pay differential became a discount was the 54th percentile. In 2018 the pay differential at the 50th percentile was -1.5% and the percentile at which the pay differential became a discount was the 45th percentile.
X-axis label | 2015 | 2016 | 2017 | 2018 |
---|---|---|---|---|
10 | 15.2576648537005 | 12.7835151377282 | 13.269528432539 | 12.7045943026467 |
20 | 11.6724671001209 | 8.59986734390565 | 8.43708965667604 | 6.76927373134981 |
30 | 7.0793540242809 | 5.11659745225668 | 4.35204031965697 | 4.15395967924728 |
40 | 4.59232623526911 | 1.24771987582037 | 1.75522618366473 | 0.81328937531524 |
50 | 0.561570931035638 | -2.66387584756632 | -0.339422654510235 | -1.53804932482671 |
60 | -1.57742622919087 | -5.01013649236277 | -2.89720244162254 | -3.1977550168694 |
70 | -4.42892743958651 | -6.01171132089111 | -5.48390041309138 | -6.24638857379503 |
80 | -6.48047986632234 | -9.43478617186912 | -8.44222571380224 | -10.4165864703472 |
90 | -12.1904569079439 | -14.0668055218354 | -13.3505918325147 | -17.2627283102544 |
Figure 4.2 shows the premia for males only for each of the four years. In 2015 the pay gap became a discount at the 36th percentile. This moved to the 16th in 2016, the 23rd in 2017 and the 13th in 2018.
X-axis label | 2015 | 2016 | 2017 | 2018 |
---|---|---|---|---|
10 | 11.0710610355705 | 6.02449856945972 | 6.11996357097904 | 3.14855038865227 |
20 | 6.47073791009516 | -1.25209523484435 | 1.19710866973839 | -2.52750983982061 |
30 | 2.67515669710561 | -4.59126024096289 | -1.71512142334388 | -2.88749167633653 |
40 | -0.796808516293934 | -6.84449570257304 | -4.20886099329694 | -5.88003556521842 |
50 | -3.81531109738048 | -8.67083426975586 | -6.98412421145718 | -9.00812780512171 |
60 | -5.97410848535372 | -12.6808160621563 | -9.07179982785022 | -10.9613819772629 |
70 | -7.61448650602358 | -13.1336973069048 | -10.9791879202138 | -12.1201811669219 |
80 | -11.511713618023 | -15.1936700212262 | -13.4545087596893 | -14.4526361270365 |
90 | -14.9813889555102 | -19.8924436573099 | -19.014495166303 | -20.1803123855346 |
Figure 4.3 shows the premia for females for the same time period. The size of the pay gap at each decile has not changed as much for females between the four years as it did for males. In 2015 the pay gap became a discount at the 74th percentile. This dropped to the 73rd percentile in 2016 and 2017 and the 68th percentile in 2018.
X-axis label | 2015 | 2016 | 2017 | 2018 |
---|---|---|---|---|
10 | 25.1947075725894 | 27.8388116010647 | 22.6175529772151 | 21.9328137398583 |
20 | 19.6738571941181 | 21.0701568182519 | 17.1869105276578 | 15.1079272399277 |
30 | 12.3782244441946 | 14.4994690640207 | 10.8824006385832 | 10.8269732962542 |
40 | 11.0488490446365 | 10.8713129529129 | 8.41540440734192 | 8.5021714330505 |
50 | 5.63293276819627 | 7.12219433338122 | 6.57726190112378 | 6.82267171659934 |
60 | 3.73817596234223 | 2.94245944751308 | 3.88351111956149 | 3.48950286710996 |
70 | 0.672249521124235 | 0.642052376066138 | 1.12629548117711 | -0.429076823693386 |
80 | -1.54789502760088 | -0.806728339442886 | -3.85377727919228 | -4.40981131352554 |
90 | -8.05687439048753 | -7.59600755549132 | -7.51280652744471 | -11.9618550422582 |
Figure 4.4 shows the premia across the earnings distribution separately for males and females for 2018. While the premium is higher for females than for males at every point throughout the earnings distribution, the difference between the two narrows at the higher end of the distribution.
X-axis label | Male | Female | Total |
---|---|---|---|
10 | 3.14855038865227 | 21.9328137398583 | 12.7045943026467 |
20 | -2.52750983982061 | 15.1079272399277 | 6.76927373134981 |
30 | -2.88749167633653 | 10.8269732962542 | 4.15395967924728 |
40 | -5.88003556521842 | 8.5021714330505 | 0.81328937531524 |
50 | -9.00812780512171 | 6.82267171659934 | -1.53804932482671 |
60 | -10.9613819772629 | 3.48950286710996 | -3.1977550168694 |
70 | -12.1201811669219 | -0.429076823693386 | -6.24638857379503 |
80 | -14.4526361270365 | -4.40981131352554 | -10.4165864703472 |
90 | -20.1803123855346 | -11.9618550422582 | -17.2627283102544 |
Figure 4.5 allows us to compare the magnitude of the pay gap across the earnings distribution for gross weekly earnings and for earnings when the pension levy is removed. On average there is a decrease of approximately 3.3 percentage points in the size of the premium when the pension levy is deducted from weekly earnings. The point on the distribution at which the pay gap becomes a discount is the 61st percentile for gross weekly earnings and the 45th percentile when the pension levy is deducted.
X-axis label | PL in | Pl out |
---|---|---|
10 | 15.4768625888289 | 12.7045943026467 |
20 | 9.58167766925413 | 6.76927373134981 |
30 | 6.92954781746002 | 4.15395967924728 |
40 | 4.02904988849579 | 0.81328937531524 |
50 | 2.00993249935899 | -1.53804932482671 |
60 | 0.300450450337708 | -3.1977550168694 |
70 | -2.34213605919128 | -6.24638857379503 |
80 | -6.38691357083812 | -10.4165864703472 |
90 | -13.2725722616233 | -17.2627283102544 |
In order to evaluate the impact the inclusion of the size of enterprise as an explanatory variable on the resulting premium, Figure 4.6 shows the premia broken down by gender for models with size of enterprise both included and excluded for 2018.
It is interesting to note that at the lower end of the earnings distribution, excluding the size of enterprise variable from the model has the effect of increasing the premium slightly, but at the higher end of the earnings distribution it has the opposite effect.
X-axis label | Males Size Incl | Males Size excl | Females Size Incl | Females Size excl | Totals Size Incl | Total Size excl |
---|---|---|---|---|---|---|
10 | 3.14855038865227 | 6.80130931856702 | 21.9328137398583 | 26.8074996790719 | 12.7045943026467 | 15.5461694963753 |
20 | -2.52750983982061 | 1.29835639383287 | 15.1079272399277 | 16.3811040743155 | 6.76927373134981 | 8.93704929466894 |
30 | -2.88749167633653 | -3.02335406212875 | 10.8269732962542 | 12.7496851579376 | 4.15395967924728 | 4.67603357472872 |
40 | -5.88003556521842 | -6.37755179411417 | 8.5021714330505 | 9.71325461283414 | 0.81328937531524 | 2.51100780052593 |
50 | -9.00812780512171 | -9.70607467712928 | 6.82267171659934 | 8.31787443844458 | -1.53804932482671 | -1.12363948050018 |
60 | -10.9613819772629 | -11.3611555931118 | 3.48950286710996 | 4.12271817780303 | -3.1977550168694 | -3.99788850283491 |
70 | -12.1201811669219 | -15.0918413402625 | -0.429076823693386 | 1.33888384133953 | -6.24638857379503 | -7.56828219916246 |
80 | -14.4526361270365 | -17.5435573540931 | -4.40981131352554 | -2.97485361505847 | -10.4165864703472 | -10.9524776702527 |
90 | -20.1803123855346 | -21.7217187390655 | -11.9618550422582 | -11.9002067664085 | -17.2627283102544 | -17.700075855458 |