4.3 Descriptive Data Analysis
4.3.6 Spain
Basic Sample Description
For Spain the sample comprises 2622 men, each with between 4 and 8 observations, with an average of 6.3 observations.31 Of the 2622 individuals, 1646 (62.8%) are initially observed in private sector employment and remain in the sample for an average of 6.4 years. There are 524 men (20.0%) originally observed in the public sector, and they are each in the sample for 6.5 years on average. The remaining 452 individuals (17.2%) are initially unemployed, and they are retained in the sample for an average of 5.9 years.
In terms of education, the Spain sample breakdown overall is 33.6% “high”, 23.0% “medium” and 43.4% “low”. However this masks vast differences in educational composition between the sectors. For the private sector the breakdown is 28.7%, 23.2% and 48.1% for high, medium and low respectively, whereas for the public sector it is 51.0%, 21.1% and 27.9%. Thus the public sector contains a much greater proportion of high educated workers and a much lower proportion of low educated workers than the private sector. The proportion in the medium education band is very close for each sector, all of the difference being in the top and bottom of the respective distributions. Additionally, the public sector workers are on average older than those in the
31
There is some sample attrition, 20.5% after 4 years and 58.9% at 8 years, which we assume to be random. Some of the attrition is a result of our sample construction rules which treat an individual as censored from the first gap in their response history (this affects 10.0% of our sample).
private sector (41.8 years old versus 38.6) and have more (potential) labour market experience (23.0 years versus 21.6). 6 8 10 12 14 16 18 20 22 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 year
Spain: unemployment rate (LFS)
17.5 20 22.5 25 27.5 30 32.5 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 year
Spain: public share of total employment (sample)
Figure 4.5: Spain: Unemployment rate, males, 1991-2001, and Sample public sector share
Figure 4.5 illustrates the evolution of the Spanish labour market over time, showing the unemployment rate for the years 1990-2001 (top panel), thus covering the years immediately preceding our sample as well as the sample years themselves, and the public sector share of total employment in our sample (bottom panel). We can see from the figure that in the years leading up to the start of the sample in 1994, the unemployment rate in Spain was rising quite sharply from around 12% in 1990 to almost 20% in 1994. During the course of the sample however, unemployment falls steadily and is down to 7.5% by 2001. The public sector share of total employment is stable at around 25-26% for the first three years of our sample before dropping slightly in 1997 (to 23%) and then continuing to fall such that is was just under 21% in 2001.
Differences in Earnings
Earnings levels. We now illustrate the public-private differences in earnings through a number of simple regressions, see Table 4.9. Again, in each case the dependent variable is the log of current gross monthly earnings. Between the sectors there are some differences in the monthly work hours distributions for full-time workers. Though the median weekly work hours are the
same for each sector (40 hours), in the private sector there are very few workers who work below the median hours and a large spike at the median with a few smaller concentrations of workers working more hours than this. For the public sector there is a smaller mass at the median and more mass at points below median hours, with much fewer working more than the median hours. Therefore we will under-estimate any positive public premium in hourly wages. Moreover, the public sector exhibits much less variance in hours (standard deviation of weekly hours is smaller by almost one hour and a half).
Column 1 of Table 4.9 shows that the raw public pay gap in our sample is very large, 26.7 log points (30.6%), in favour of the public sector. This is unsurprising given the great differences in educational, and to a lesser extent age, composition of the respective workforces. Controlling for education and a quadratic in experience (column 2), the premium is reduced to 9.7 log points (10.2%). Allowing the public effect to differ according to education band and experience (column 3), we find no different effect of public sector employment dependent on education but the premium does vary with experience, from 27.0 log points (30.9%) prior to any labour market experience, to a low of 8.5 log points (8.9%) at 28 years experience. Estimating with individual-level fixed effects and controlling for just a quadratic in experience (column 4), the public premium is not significantly different to zero.32 However, when we add to this fixed effects specification the interaction terms between the public dummy and experience and its square, the public premium is significant and estimated to be 16.2 log points (17.6%). Believing this latter fixed effects specification, we conclude that the returns to experience are consistently lower in the public sector, the initially large public premium tracing a U-shape in experience, turning negative after 25 years of experience and remaining negative for almost all the rest of a working lifetime. The estimated minimum of the public premium is−1.7 log points (−1.7%) and comes at 37 years of experience.
Earnings dispersion. There is a greater degree of wage compression in the public sector, illustrated in the standard deviation of log earnings, which is 0.395 in the public sector compared to 0.427 in the private sector. The 90:10 percentile ratios in raw earnings, 2.803 and 2.965 for the public and private sectors respectively, support this conclusion, as do the corresponding ratios
32
The reported fixed-effects regressions use the within-estimator. The first differenced estimate is very similar for Spain specification 4.
controlling for age and education: 2.354 for the public sector and 2.706 for the private sector.
Earnings mobility. The regression results and analysis of the earnings distributions show that there are differences between the public and private sectors in terms of earnings levels and cross-sectional distributions. Furthermore, the extent of earnings mobility differs between the two sectors, as illustrated by Table 4.10 which shows the transition matrices between un- conditional log earnings quintiles from one year to the next for the public and private sectors respectively.33 These matrices show that persistence in earnings rank is greater amongst indi- viduals continuously employed in the public sector compared to those continuously employed in the private sector; and this is especially true at the lower end of each distribution. Moreover, the lower panel of the Table, which shows the transitions for log earnings residual rank after conditioning on education and a quadratic in (potential) labour market experience, illustrate the same pattern of greater earnings persistence in the public sector.
Since the earnings distribution is less compressed in the private than in the public sector, transitions between quintiles in the private sector represent greater rises (falls) in earnings than similar transitions in the public sector, which further emphasizes the difference in earnings mobility between the sectors. Comparing the average one-lag auto-covariance of normalised log income, after controlling for education and a quadratic in (potential) labour market experience, the greater auto-covariance for individuals employed for consecutive periods in the public sector (0.890 versus 0.751, the corresponding figure for the private sector) adds to the picture of more persistent earnings in the public sector.
Differences in Job Mobility
The transition matrix below shows the changes in employment sector between one wave and the next, with rows referring to sector in yeart-1, columns to sector in year t:
Private Public Unemp. Private 91.8 1.6 6.5 Public 7.4 89.9 2.5 Unemp. 39.2 5.4 55.3 33
From the table it is apparent that movements directly from private sector employment to public sector employment in the following year is rare, only 1.6% of those in the private sector in year t-1 are employed in the public sector in year t, however movements from public-to- private happen with greater frequency (7.4%). The annual transition rate into unemployment is, for the public sector, less than half what it is for the private sector (2.5% public versus 6.5% private) suggesting greater job security in the public sector. This is further underscored by the differing probabilities of ever being observed in unemployment depending on the sector in which an individual is first observed. Of those first observed in the private sector, 25.6% are ever subsequently observed unemployed, while for individuals first observed in the public sector, the figure is only 8.2%.
Of the unemployed in year t-1, 55.3% remain unemployed the following year, with 39.2% exiting to private sector employment, and the other 5.4% finding employment in year t in the public sector. There are 917 individuals (35.0% of the sample) who are ever observed unemployed during the time-span of the sample. Of these, 28.8% report being unemployed in three or more consecutive interviews during the course of their observations. These “long-term” unemployed have an 18.9% chance of finding employment in the next period, while for the “short-term” unemployed the average annual re-employment rate is much higher at 73.0%. Thus though unemployment persistence is quite low overall, it appears to be concentrated on the “long-term” unemployed.
There is also evidence of public sector attachment, in the differing public sector re-employment probabilities between the unemployed whose most recent observed employment was in the public sector and those whose most recent employment was in the private sector. For the former group, the probability of gaining a public sector job by the next year is 17.2% while for the latter group it is only 3.4%.