Econometric Analysis of the Public-Private Sector Pay Differential 2022

Key Findings

• Between 2019-2022 there was no clear trend in the public-private pay differential. Due to the impacts of COVID-19 on the labour market and earnings from 2020-2022, it is too early to see if reductions in the public-private pay differential between 2018-2020 are a definitive trend.

• Including and excluding the size of enterprise from the models has a significant effect on the pay differential. When firm size is included, public sector employees are worse off in terms of gross pay with discount ranges of -4.4% to -6.6% in 2022.

• When firm size is excluded, public sector employees are slightly better off in terms of gross pay in comparison with private sector employees, with differential ranges from a premium of 0.9% to a discount of -1.4% in 2022.

• Trends in the public-private pay differential show that public sector employees on higher incomes are worse off in terms of gross pay in comparison with private sector employees, this is known as the public sector discount.

• This trend reverses for those on lower incomes in the public sector who are better off in terms of gross pay in comparison with private sector employees, this is known as the public sector premium (See Figure 4.1).

• The size of the pay differential in the public sector was higher for females than for males i.e. females in the public sector have a higher pay differential than males in the public sector when compared to their private sector counterparts (See Figure 4.4).

Statistician's Comment

The Central Statistics Office (CSO) has today (07 May 2024) issued Econometric Analysis of the Public-Private Sector Pay Differential for 2019-2022.

Commenting on the data, Darragh Turner, Statistician in the Earnings Analysis Division, said:

"Due to the complex nature of measuring the public-private pay differential, four different estimates are presented in this paper. In the international literature there is no clear uniform method for producing the public-private pay gap, hence, there are a number of methods incorporated in this paper. This ranges from the models used, in this case Ordinary Least Squares (OLS) Regression and Quantile Regression (See Editors Note below for definitions), and also what is included and excluded in the model specifications i.e. size of enterprise (enterprises with 100 employees or more, or those with fewer than 100 employees) and additional superannuation contribution.

OLS Regression Results

Results from the OLS Regression model show a public-private sector pay differential ranging from 0.2% in 2019 to 0.9% in 2022, for the model which excludes size of enterprise as a determining factor. Results for the OLS model which deducts the additional superannuation contribution and includes size of enterprise shows a pay differential ranging from -7.8% to -6.6% (See Table 4.1).

Including and excluding size of enterprise from the models has a significant effect on the pay differential. When firm size is included, the public sector discount ranges from -4.4% to -6.6% in 2022. With firm size excluded, the public sector differential ranges from a premium of 0.9% to a discount of -1.4% in 2022.

Between 2019-2022 there was no clear trend in the public-private pay differential. However, there does appear to be a levelling off in the public-private sector pay differential from 2016 with slight reductions from 2018-2020. Figures have generally remained stable over this period with minor fluctuations. Due to the impacts of COVID-19 on the labour market and earnings from 2020-2022, it is too early to see if reductions in the public-private pay differential between 2018-2020 are a definitive trend during this period.

When comparing the public and private sector for 2022, the pay differential for male employees in the public sector ranged from a discount of -4.5% to a discount of -11.0% depending on the specification used in the model.

The corresponding differential for females showed that female workers in the public sector had a pay differential ranging from a premium of 6.7% to a discount of -2.2% depending on the specification used in the model. The size of the pay differential in the public sector was higher for females than for males i.e. females in the public sector have a higher pay differential than males in the public sector when compared with their private sector counterparts (See Figure 4.4).

Quantile Regression Results

Summary results from the Quantile Regression model show a public-private sector pay differential in 2022 ranging from 19.8% at the 10th percentile to -11.9% at the 90th percentile for the model which excludes size of enterprise as a determining factor (See Figure 4.1 and Table 8.4). The corresponding model which excludes the additional superannuation contribution and includes size shows a pay differential in 2022 ranging from 11.0% at the 10th percentile to -17.6% at the 90th percentile (See Table 8.8).

Trends in the public-private pay differential show that public sector employees at the upper end of the earnings distribution are worse off in terms of gross pay in comparison with private sector employees, this is known as the public sector discount. This trend reverses at the lower end of the earnings distribution where public sector employees are better off in terms of gross pay in comparison with private sector employees, this is known as the public sector premium (See Figure 4.1)."

Editor's Note

This research paper presents an econometric analysis of the public-private sector pay differential for the period 2019 to 2022 and has been prepared in response to user needs to inform discussions relating to the composition of earnings.

The methodology employed in this analysis is the same as that used by the Central Statistics Office (CSO) to produce the analysis for years 2011-2014 and 2015-2018. See References Chapter. A combination of available survey data and administrative data sources has been used. The sources used are the CSO's Labour Force Survey (LFS) and the Earnings Analysis using Administrative Data Sources (EAADS).

The methods used in these analyses are: Ordinary Least Squares Regression (OLS); and Quantile Regression. For each of these methods, results based on a range of specifications are presented.

OLS regression is a statistical method used to estimate the relationship between one or more independent variables (e.g. Occupation, education, nationality, age, public-private status, hours worked, overtime hours worked, length of service, trade union membership, supervisor role) and a dependent variable (e.g. weekly gross pay).

Quantile regression is similar to OLS regression, but instead of predicting the mean of the dependent variable, it predicts specified quantiles (percentiles). In this case, at every 10th percentile. Quantile regression is particularly useful when the rate of change in a variable differs across levels of a factor. It provides a more complete view of possible causal relationships between variables in a regression model.

This research paper provides a range of results using these two methods. See Methodology and Data Sources Chapter.