28 As proposed by Christofides et al. (2013), there exists glass ceiling effect if gap at 90th percentile is at least 2 percentage points larger than the gap for a reference percentile. Likewise, sticky floor effect is present when the gap at 10 th percentile is 2 percentage points larger than the reference percentile gap size. Due to the fact that IHS transformed income is linear around its origin, the gap for very low in- comes should be interpreted in monetary values, while gaps for larger incomes can be interpreted, roughly speaking, in percentages (to be more precise, as in case of logarithmic transformation). The former is mostly observed at 10 th percentile of the distribution. For this reason, to show existence of sticky floor effects, instead of 10 th percentile, as described by Christofides et al. (2013), 20th percentile will be compared to reference gap. For the reference, median employment income gap is taken. These effects can be easily seen from the shape of the unconditional employment income gap on Figure B.2. U-shaped raw gap implies that both effects could be present in a country, then this country can be further examined whether it satisfies the suggested definition of the effects. On the contrary, inverse U-shaped raw gap shows that neither of these effects are present in a country. The increasing and de- creasing graphs can be a good indication of the presence of glass ceiling and sticky floor effects, respectively. However, there are countries, where the unconditional gap has complex shape and no prior assumption can be made. Table 5 shows whether there is the evidence of either sticky floor or glass ceiling effects. It can be observed that in Belgium, Czechia, France, Greece, and Norway both effects are present in both age groups. The presence of these effects in Greece is in line with the findings of Olivetti and Petrongolo (2008), who showed that labor market is mostly populated by highly skilled female workers (positive selection). However, the results partially conform with the findings of Christofides et al. (2013), who showed that neither of these effects is present in Greece and Spain. In Finland and the UK sticky floor and glass ceiling effects are present in the age group of 25-44, while only the latter is observed in the older age group. On the contrary, in Slovakia and Sweden both effects are reported for the age group 45-65, while only sticky floor is observed in the younger age group. In Austria, Germany, and the Netherlands only sticky floor effect is observed in both age groups. In contrast, only glass ceiling effect is present in Bulgaria and Lithuania for both age categories. In Croatia there is the evidence of only sticky floors effect in age group 25-44. Similarly, in Estonia only glass ceiling is observed in the age group 45-65. On the contrary, in Poland, Portu- gal, Romania, and Serbia only glass ceiling is reported for the age group of 25-44. In Denmark, glass ceiling effect is observed in the younger age group, whereas the evidence of both effects is reported for the older age group. The data provides the evidence of sticky floor and glass ceiling effects in the age group of 25-44 and 45-65, respectively, in Latvia and Slovenia. The evidence of sticky floor effect is found in Spain for the age group 45-65. Hungary is the only country, where both sticky floor and glass ceiling effects are observed in the age group 25-44, while only sticky floor evidence is provided for the age group 45-65. 5. Results 29 TABLE 5: Sticky floor and glass ceiling effects over age groups. Age Group 25 − 44 45 − 65 Sticky floor Glass ceiling Sticky floor Glass ceiling Austria Yes Yes Belgium Yes Yes Yes Yes Bulgaria Yes Yes Croatia Yes Czechia Yes Yes Yes Yes Denmark Yes Yes Yes Estonia Yes Finland Yes Yes Yes France Yes Yes Yes Yes Germany Yes Yes Greece Yes Yes Yes Yes Hungary Yes Yes Yes Latvia Yes Yes Lithuania Yes Yes Netherlands Yes Yes Norway Yes Yes Yes Yes Poland Yes Portugal Yes Romania Yes Serbia Yes Yes Slovakia Yes Yes Yes Slovenia Yes Yes Spain Yes Sweden Yes Yes Yes UK Yes Yes Yes Note: Sticky floors effect is present if the total gap at the 20 th percentile is at least 2 percentage points larger than the gap at 50th percentile. Note: Glass ceiling effect is present if the total gap at the 90 th percentile is at least 2 percentage points larger than the gap at 50th percentile. Source: author’s calculation from EU-SILC 2016. 5.3 Results for the gap in private transfers and capital income In this subsection, principal findings of the gender gap in private transfers and cap- ital income are presented. As shown in Figure 1, the private transfers and capital income constitute significant portion of the total income in some countries (for ex- ample, France, Greece, Spain, the UK), while its share is relatively low in other coun- tries (for example, Serbia, Slovakia, Slovenia). In the post-communist countries, the relatively short time to accumulate wealth could possibly explain why this income category has relatively low share (Meriküll et al. (2018)). The gender gap in this income category is analyzed in the similar manner as for the total income: first each 30 age group is analyzed across these countries, and then its development over the life- cycle is investigated. Figure 4 reports results for all countries, except Romania, which has been ex- cluded due to data limitations. For the youngest age group, it can be observed that most of the estimates are statistically insignificant. Among those, which are signif- icant at the given confidence level, France has the largest unconditional (0.838 log points) and conditional (0.986 log poins) gaps in favour of men. However, the dif- ferences in observed characteristics benefit women more than men. On the contrary, in Hungary, both unconditional and conditional gaps are in favour of women (1.116 and 0.882 log points, respectively16). What is even more striking is that the observed characteristics benefit women more as well17. FIGURE 4: Median gap in private transfers and capital income for all countries. −.5 0 .5 1 Gap size <25 25−44 45−65
65 Age group Austria −1 −.5 0 .5 1 Gap size <25 25−44 45−65 65 Age group Belgium −.5 0 .5 1 1.5 Gap size <25 25−44 45−65 65 Age group Bulgaria −1 0 1 2 Gap size <25 25−44 45−65 65 Age group Croatia −1 −.5 0 .5 1 Gap size <25 25−44 45−65 65 Age group Czechia −2 −1 0 1 2 3 Gap size <25 25−44 45−65 65 Age group Denmark −2 −1 0 1 2 Gap size <25 25−44 45−65 65 Age group Estonia −.5 0 .5 Gap size <25 25−44 45−65 65 Age group Finland −.5 0 .5 1 1.5 Gap size <25 25−44 45−65 65 Age group France −.4 −.2 0 .2 .4 Gap size <25 25−44 45−65 65 Age group Germany −.8−.6−.4−.2 0 .2 Gap size <25 25−44 45−65 65 Age group Greece −1.5 −1 −.5 0 Gap size <25 25−44 45−65 65 Age group Hungary −4 −2 0 2 4 Gap size <25 25−44 45−65 65 Age group Latvia −10 −5 0 5 Gap size <25 25−44 45−65 65 Age group Lithuania −.5 0 .5 1 1.5 Gap size <25 25−44 45−65 65 Age group Netherlands −.4−.2 0 .2 .4 .6 Gap size <25 25−44 45−65 65 Age group Norway −1 −.5 0 .5 1 Gap size <25 25−44 45−65 65 Age group Poland −1 −.5 0 .5 Gap size <25 25−44 45−65 65 Age group Portugal −3 −2 −1 0 1 Gap size <25 25−44 45−65 65 Age group Serbia −1.5 −1 −.5 0 .5 Gap size <25 25−44 45−65 65 Age group Slovakia −2 −1 0 1 Gap size <25 25−44 45−65 65 Age group Slovenia −1 −.5 0 .5 1 1.5 Gap size <25 25−44 45−65 65 Age group Spain −.5 0 .5 1 1.5 Gap size <25 25−44 45−65 65 Age group Sweden −1.5−1 −.5 0 .5 1 Gap size <25 25−44 45−65 65 Age group UK Raw gap Unexplained gap Note: confidence intervals are shown for 90% confidence level. Note: Romania is excluded due to data limitations. Source: author’s calculation from EU-SILC 2016. 16Please, note that gaps are reported as positive values, since they are in favour of women. 17In the Figure 4, Lithuania has also somewhat large median gap benefiting women, however, as described in the previous subsection, such large gap is an indicator that the differential is given in absolute values. Thus, it is difficult to say decisively that in Lithuania there is a larger gap favouring women than in Hungary.
- Results 31 In the following age group of 25-44, the unconditional median gap in private transfers and capital income is systematically in favour of women and, occasionally, in favour of men (e.g., Denmark, France, and Lithuania), however, rarely statistically significant. The differences in the observed characteristics always favour women, except in Serbia. As for the next age group of 45-65, the gap predominantly favours women. In France, Norway, and Portugal the observed characteristics benefit men more than women. The largest unconditional median gap is reported for Denmark, however, it is statistically insignificant. In Croatia there is the largest statistically sig- nificant unconditional median gap in favour of women (0.668 log points). Among the people over age 65, the number of countries, where the unconditional median gap favours men, increases: in 11 out of 24 countries the median raw gap is positive. However, out of these 11 countries the median gap is statistically signif- icant only in Austria, Belgium, Norway, Sweden, and the UK. The largest median gap, which favours men, is reported for Belgium (0.84 log points), followed by the UK (0.773 log points). On the contrary, in Slovakia there is the largest gap in favour of women (0.554 log points). There are countries, where the median gap peaks for the youngest and the old- est individuals. For example, in Austria, the Netherlands, Norway, and Sweden, the gap has U-shaped pattern and favours men for individuals below age 25 and over age 65. On the contrary, there are countries, where the same pattern occurs, however, in favour of women. For example, in Greece, Lithuania, Serbia, and Slo- vakia, in the youngest and oldest age groups there is the largest median gap, which favour women. In Bulgaria, Croatia, Denmark, France, and Spain the unexplained median gap is favour of men in the youngest age group, then gradually decreases and favours women for the eldest individuals. 5.4 Results for the gap in public transfers The final income category that is analyzed in this paper is public transfers. The de- tailed decomposition of the gap is provided in Tables A.19, A.20, A.21 and A.22 for each age group, respectively. The information regarding the median gap in public transfers is shown in Figure 5. In the youngest age group, the unconditional median gap is observed to be pre- dominantly in favour of women, and occasionally in favor of men. In the countries, where the raw gap favours men (Austria, Bulgaria, Croatia, Greece, Latvia, Poland, and Serbia) the gap is not statistically significant. On the other hand, in 9 out of 18 countries (Denmark, Estonia, Hungary, the Netherlands, Norway, Slovakia, Slove- nia, Sweden, and the UK), the gap, favouring women, is statistically significant. The largest gap is reported in the Netherlands, where the gap is mostly unexplained by the differences in the observed characteristics (-0.672 out of -0.832 log points). However, in Denmark, the country with the second largest median gap in favour of women, only a little portion of the raw gap is unexplained (0.144 out of 0.82 log points). In the majority of countries, where the unconditional median gap in pub- lic transfers favours women, both explained and unexplained gaps are in favour of women as well. In Belgium, Germany, Portugal, and Spain the observed character- istics benefit men more.