question in description

Ireland’s Recession and the Immigrant/Native Earnings Gap

Alan Barrett[footnoteRef:1], Adele Bergin, Elish Kelly and Séamus McGuinness [1: Corresponding author: Alan Barrett, e-mail. [email protected]. We would like to acknowledge the helpful comments of participants at a workshop in CEU, Budapest in October 2012. ]

Economic and Social Research Institute

December 2012

Abstract

In the mid-2000s, Ireland experienced a large inflow of immigrants owing to favourable economic conditions and its decision to allow full access to its labour market to citizens of the EU’s New Member States in May 2004. This chapter begins by reviewing research on the labour market outcomes of immigrants in Ireland and the effect of immigration on the welfare state: this research relates mainly to the pre-crisis period. Beginning in 2008, Ireland experienced a sharp contraction in economic activity - between 2007 Q4 and 2011 Q1 real GNP fell by 12 per cent. Given the changed economic climate, and evidence that the recession has had a severe impact on immigrant employment in Ireland, it is interesting to ask how immigrant earnings have developed over the crisis relative to native earnings. The results in this chapter show that there has been a large increase in the raw immigrant\native wage gap over the crisis, but that the wage gap has fallen slightly when we control for education, tenure, etc. A decomposition analysis indicates that most of the increase in the raw wage gap is driven by compositional changes, in particular, a fall in the share of immigrants with degrees and the share of immigrants in the relatively well-paid public sector occupations.

1. Introduction

Between the mid 1990s and the present time, the Irish economy has experienced periods of growth and contraction which were large by international and historic standards. In Figure 1, we show rates of growth in real GDP and real GNP for the period 1996 to 2011 and the contrasting performance of the Irish economy over the period is clear. In the mid-to-late 1990s, the Irish economy grew at annual rates in the region of 10 percent. While growth moderated in the early years of the 2000s, with annual rates of growth around 5 percent, Ireland’s economic performance still looked remarkably healthy. However, when the global crisis of 2007/8 emerged, the Irish economy was found to be extremely vulnerable.

Figure 1: Rates of growth in Ireland’s real GDP and real GNP, 1996 to 2011

Source: Central Statistics Office

The international financial collapse had a severe affect on Ireland due partly to the existence of a property price bubble and excessive bank lending into the property sector. As shown in Figure 1, Ireland’s real GDP fell in each year between 2008 and 2010, by 2 percent, 5 percent and 1 percent respectively. The decline was even more pronounced when viewed in terms of GNP, with the economy contracting by 8 percent in 2009 alone based on this measure. The economic collapse led to a severe contraction in tax revenues. When combined with the banking-related liabilities which the state accrued as a result of the banking guarantee, a public finance crisis emerged and this led to the EU/IMF bailout in 2010.

The dramatic movements in Ireland’s economy have been mirrored in the migratory movements into and out of Ireland. In Figure 2, we present rates of net inflows (in thousands) for the period 1987 to 2012. Although the height of the economic boom occurred in the late 1990s, the surge in inward migration is observed in the period after 2004. The main reason for this was the enlargement of the EU in 2004 and the fact that Ireland was among only three countries which allowed full access to its labour market to the citizens of the New Member States (NMS) from 1 May 2004. In 2007, net inward migration peaked at over 100,000. The population of Ireland was measured at 4.2 million in 2006 so the net inflow of 100,000 represented 2.4 percent of the population.

It can also be seen in Figure 2 that the economic crisis has led to a reversal of net migration. In the year ending April 2012, the net outflow was over 34,000. In absolute numbers, this means that the rate of net outflow is now approaching the rate of the late 1980s when Ireland last experienced large population net outflows.

Figure 2: Net inflows into Ireland (in thousands), 1987 to 2012

Source: Central Statistics Office

In this chapter, we will take a closer look at migration to and from Ireland over the recent past. In Section 2, we will discuss the research findings that arose from the work conducted during the boom. As some of this work has been discussed in detail elsewhere, for example in Barrett (2010), we will focus on the broad findings here. In Section 3, we will present some findings on the employment outcomes of immigrants in Ireland over the crisis. It will be seen in that section that immigrants suffered higher rates of job losses over the crisis relative to natives. In Section 4, we present new findings on the immigrant/native wage gap over the crisis. While we know from previous work that the employment of immigrants fell over the crisis, we do not know what happened to their earnings relative to natives. For this reason, Section 4 contains an analysis of the evolution of the immigrant-native wage gap between 2006 and 2009, the year of the deepest contraction.

2. Immigration into Ireland during the boom

During the mid-to-late 2000s, the migration-related research agenda pursed by economists in Ireland followed that which had been pursued by economists in traditional immigrant receiving countries. As a result, the key questions that were addressed were as follows: what was the immigrant/native wage gap; did immigrants assimilate in terms of labour market outcomes; what were the impacts of immigrants on native wages, employment and the public finances.

The main papers on the immigrant/native wage gap were Barrett and McCarthy (2007) and Barrett et al. (2012). Both papers revealed the familiar finding that immigrants generally earned less than natives, even when account was taken of characteristics such as age and education. Based on data from the Irish component of the EU Survey on Income and Living Conditions (EU-SILC) 2004, Barrett and McCarthy (2007) found that immigrants earned 18 percent less than comparable natives. However, this figure concealed large differences across immigrant groups. There was no difference in the earnings of natives and immigrants from English-speaking countries. However, there was a 45 percent difference between natives and immigrants from the NMS of the EU.

Barrett et al. (2012) used a much larger dataset from 2006 (the National Employment Survey) to look again at the immigrant/native wage gap and to expand the analysis. They also found a wage gap between natives and immigrants from the NMS, although at 18 percent, the estimate was lower than that in Barrett and McCarthy. Barrett et al. (2012) also showed that the wage gap differed across the earnings distribution. Using quantile regressions, they found no difference in earnings at the lower end of the distribution but large differences at the higher end. They interpreted these findings as suggesting that the immigrant wage disadvantage was more likely to result from a failure to achieve comparable rates of return on human capital for higher skilled immigrants as opposed to discrimination and exploitation at the lower end of the labour market.

Barrett and Duffy (2008) was the only paper which directly addressed the question of assimilation in the labour market. In this paper, the authors used data from the quarterly national labour force survey and explored whether immigrants who had been longer in Ireland had better occupational outcomes. The authors found evidence of an “occupational gap” for immigrants – while immigrants in Ireland were relatively well-educated, the occupations which they were employed in did not fully reflect this. No evidence was found for a relationship between the size of the occupational gap and time spent in Ireland. From this, Barrett and Duffy (2008) concluded that they were unable to find evidence of labour market assimilation.

The most comprehensive study of the effects of immigration on the Irish economy was contained in Barrett et al. (2006). In this paper, the authors used a structural model of the Irish labour market and the macro-economy to simulate the impact of immigration on wages, employment and output, along with a range of other variables. The study showed that immigration had increased GDP and GDP per capita during the late 1990s and early 2000s. The mechanism through which this occurred was noteworthy. Immigration was shown to have helped to dampen wage pressures. This was an important role in the Irish economy in the early 2000s because costs were rising and competitiveness was falling. Given the importance of external demand to a small open economy such as Ireland, competiveness is a key driver of growth. Without immigration on the scale seen during the boom, it is argued that wages would have risen, thereby choking off labour demand and curtailing economic growth.

The analysis of Barrett et al. (2006) and others may have contributed to a generally favourable assessment of the impacts on immigration in Ireland. A set of results which added to the favourable assessment were found in Barrett and McCarthy (2007), Barrett and McCarthy (2008) and Barrett et al. (forthcoming 2013) where the research question related to welfare receipt on the part of immigrants. In many countries in recent times, negative sentiment towards immigrants appears to be voiced, in part, through expressions of concern about the relatively intensive use of welfare payments and services on the part of immigrants. The results in the studies listed above showed that immigrants in Ireland were less likely to be in receipt of welfare payments. This observation was likely to have been the result of policy, at least to some degree. At the time of EU enlargement, the Irish government created a residency requirement for the receipt of welfare payments which essentially meant that people would have to have been resident in Ireland for at least two years before eligibility for welfare applied.

In summary, the research that was conducted on immigration into Ireland during the boom produced a picture of immigration that was positive for Ireland. Even though immigrants may have dampened wage pressures, wages were still rising over this period and unemployment was low. The economy appeared to be absorbing immigrants with minimal disruption. The only concerns related to the immigrants themselves, especially those for the NMS. Their wages were well below those of comparable natives, partly because of a failure to access occupations which fully utilised their human capital. This would not be a concern if any initial disadvantages in the labour market reduced with time spent in Ireland, but no evidence existed that assimilation was occurring.

3. What happened to immigrant employment during the crisis?

As the economic crisis began to take hold in Ireland in 2007/8, Ireland’s immigrant population, especially those from the NMS, was made up of many recently arrived immigrants. As shown in Figure 2 above, immigration had surged in the mid 2000s and so while some immigrants had arrived in the 1990s and before, there was a huge group of new arrivals. We also know from the research on labour market outcomes that many of the immigrants from the NMS were in lower earning and less-skilled jobs. These features of the immigrant population in Ireland suggested that they might be vulnerable in an economic downturn and this turned out to be the case.

In Figure 3, we show annual rates of employment change in recent years for Irish nationals and for immigrants from the NMS. Looking firstly at the earlier part of the figure, the phenomenal growth in immigrant employment up to Q3 2007 can be seen. At that time, the annual rate of employment growth for this group was almost 40 per cent. By contrast, there was barely any growth in the employment of Irish natives, thereby demonstrating the importance of immigration to growth in Ireland at that time.

As shown in Figure 1 above, 2009 saw the most severe contraction in GDP and this is reflected in Figure 3 in terms of employment falls. However, there was a large difference between the rates of employment fall for Irish natives and for immigrants from the NMS. Native employment fell by 8 percent in the year to Q3 2009; the corresponding figure for the NMS was 18 percent. For all immigrants, the rate of employment loss was 12 percent. Hence, the rate of employment loss generally among immigrants exceeded that of natives but the rate of loss was highest for the immigrants from the NMS. In 2010, the rate of employment loss eased but remained high, as did the differential between natives and immigrants from the NMS. For Irish natives, employment loss was 3 percent to Q3 2012; the corresponding number for MNS immigrants was 11 percent. For immigrants in general, the rate of employment loss to Q3 2012 was 8 percent.

The numbers from the NMS who were employed in Ireland peaked at 175,000 in Q1 2008. This represented over 8 percent of those employed. By Q1 2011, the number employed had fallen to 121,000, a fall of over 30 percent. As a proportion of those employed, the Q1 2011 figure was 6.6 percent.

Figure 3: Annual rates of employment change, Irish and EU NMS nationals, Q3 2007 to Q3 2012

Source: Central Statistics Office

This huge fall in the employment of immigrants generally, but the NMS immigrants in particular, was presented by Barrett and Kelly (2012). In that paper, they also used micro-data from 2008 and 2009 to explore whether the employment fall was truly the result of immigrants status as opposed other characteristics that would be correlated with being an immigrant such as age and sector of employment. Their results confirmed that immigrants were disproportionately impacted upon by the recession with immigrants from the NMS being most severely affected.

The employment losses experienced by both immigrants and natives, as shown in Figure 3, are clearly related to the net outflows shown in Figure 2 above. In that figure, we gave inflow and outflow numbers which capture all such movements without differentiating by nationality. For the year ending April 2012, the total net outflow was 34,400. Of this, just over 25,000 was a net outflow of Irish nationals with a further 4,400 being the net outflow of NMS nationals. This shows how a large proportion of the current outflow is made up of Irish nationals. However, relative to their respective populations, the outflow of NMS nationals has been large. Since 2008, when the number of nationals from the NMS living in Ireland peaked at 248,000, the number of NMS nationals has fallen to 229,000 (a fall of almost 8 percent).

To summarise, Ireland’s economic crisis led to large employment losses for both nationals and non-nationals but the job losses among the NMS nationals were particularly acute. The job losses, and the general weakness in the labour market, have led to migratory outflows. In absolute numbers, the net outflow of Irish nationals is now the larger component of the net outflow. However, relative to their populations resident in Ireland, the outflows are impacting most strongly on NMS immigrants.

4. What happened to immigrant earnings during the crisis?

In this section we explore what happened to the immigrant\native wage disadvantage over the crisis. Barrett and Kelly (2012) have shown that employment losses over the recession were higher among immigrants than natives, even after controlling for relevant variables such as age and education. This might indicate that labour demand for immigrants relative to natives fell and hence there may have been more downwards pressure on immigrant wages relative to native wages as a consequence of the recession. As a result, we might expect the immigrant\native wage gap to have increased over the course of the recession. In addition, we know that the wage gap in 2006 was higher for more skilled immigrants (Barrett et al., 2012). If employment losses were concentrated among high-skilled immigrants, for whom the wage gap was higher, then this would tend to reduce the wage gap. We investigate the changes in the immigrant\native wage disadvantage over the recession using standard OLS wage models including variables for migrant status. We also further examine the change in the raw wage gap by decomposing the change into observable components i.e. changes in composition or in the return(s) to particular attributes and unobservable components to assess what is driving the change in the wage disadvantage over the period.

4.1 Data and Methods

The data used in this analysis comes from the October 2006 and October 2009 waves of the National Employment Survey (NES). The NES is a cross-sectional employer-employee linked workplace survey covering both the public and private sectors,[footnoteRef:2] which is carried out by the Central Statistics Office (CSO).[footnoteRef:3] The employer sample is drawn from the CSO’s Central Business Register and those firms that are chosen are asked to select a systematic sample of employees from their payrolls: a total of 8,383 firms were surveyed in 2006 employing 68,427 individuals, while the corresponding firm and employee figures for the 2009 survey were 9,108 and 102,208 respectively. From these samples, 4,209 (4,753) firms and 51,252 (67,907) employees completed their respective questionnaires in 2006 (2009), which gives a response rate of just over 50 (52) per cent from firms and 75 (66) per cent from employees. In this analysis, we focus on employees aged 15 and over, which gives us a final sample of 50,130 employees in 2006 and 66,122 in 2009.[footnoteRef:4] To ensure that our results are representative of the working age employee population, we apply cross-sectional weights. [2: Agriculture, forestry and fishing is the only sector that is excluded from the NES. As it is a work-place survey, the self-employed are excluded as well. ] [3: While the NES consists of enterprises with three or more employees, the results are calibrated to the Quarterly National Household Survey (QNHS) employment data for employees (excluding agriculture, forestry and fishing), which covers all employees.] [4: Characteristic information on the sample used in the study, broken down by nationality, is available from the authors on request.]

The NES employer questionnaire captures data on employee earnings, hours worked, firm size, occupation, industry, pay agreements and company training. In the employee questionnaire, gender, age, nationality, educational attainment, employment history, professional body membership, trade union membership and other job-related characteristic (e.g. shift-work, supervisory role, flexi-time, etc.) data are gathered. The earnings information collected in the NES represents the gross monthly amount payable by the organisation to its employees, and the reference month is October.[footnoteRef:5] [5: The gross monthly measure, which is earnings before the deduction of tax, social insurance contributions and superannuation, includes i) normal wages, salaries and overtime; ii) taxable allowances, regular bonuses and commissions; and iii) holiday or sick pay for the period in question, while it excludes i) employer’s PRSI, ii) redundancy payments and iii) back pay.]

Table 1 shows average gross hourly earnings[footnoteRef:6] for Irish employees and immigrants in 2006 and 2009, along with the hourly earnings of various immigrant groups; specifically, United Kingdom (UK), EU-15 (excluding Ireland and UK), EU-10 and EU-2 (NMS), Non-EU English speaking and Non-EU Non-English speaking. Prior to the current recession, natives earned €21.55 per hour in 2006, while immigrants earned less at €19.52. However, this immigrant average masks considerable variation amongst the different nationality groups. In particular, NMS employees recorded the lowest hourly earnings (€12.10 for EU-10 and €14.26 for EU-2), while Non-EU non-English speaking (€23.15) and UK nationals (€22.99) had the highest. Contrary to what one might expect, natives’ hourly earnings grew over the course of the recession, increasing by 7.5 per cent to €23.16 in 2009, while immigrants’ wages fell by 8.4 per cent to €17.89. Again, this immigrant average conceals differences between the nationality groupings with UK (€23.69), EU-15 (€22.44), EU-10 (€12.78) and EU-2 (€15.70) employees’ hourly earnings also growing over the period, while Non-EU workers earnings fell. While this table illustrates changes in hourly earnings for natives and various immigrant groups between 2006 and 2009, the econometric analysis focuses on explaining changes in hourly earnings between Irish and immigrant workers as a whole. [6: The NES data contains an hourly earnings variable, which has been derived from the gross monthly measure.]

Table 1: Average Earnings per hour (€) for Natives and Immigrants

Note: 2009 wage data is expressed in 2006 prices.

Figure 4 shows the quarterly profile for output and demonstrates clearly that economic activity collapsed over the period 2008 to 2010. By 2011 Q1 real GDP (GNP) was 10 (12) per cent below its 2007 Q4 level. The two vertical lines in Figure 4 highlight the period covered by the NES data used in this study which, we argue, encapsulates the bulk of the downturn in economic activity and, therefore, should capture most of the adjustment that took place with respect to earnings.

Figure 4: Quarterly GDP and GNP

Note: Both series are seasonally adjusted and expressed in constant market prices. Source: Quarterly National Accounts, CSO.

With respect to the methodology used, we decompose the change in the immigrant\native wage gap between 2006 and 2009 using the Juhn-Murphy-Pierce methodology (JMP, 1993). This framework allows us to deconstruct changes in the wage gap over time into changes in the distribution of workers’ observable characteristics (a quantity effect), variations in the returns to observed characteristics (a price effect) and changes in the distribution of residuals (an unobserved effect).

We can write the standard Mincerian wage equations for natives and immigrants as follows:

where is gross hourly earnings, denotes human capital, job and industry characteristics, is a vector of coefficients, is the errror term, is for natives and denotes immigrants. If we estimate these equations using OLS we can write the average wage difference between natives and immigrants in, say 2006, as:

where , and

Then we can write the change in the native\immigrant wage gap between 2006 and 2009 as:

The first term in (4) measures the change in the immigrant\native wage gap between 2006 and 2009 that is due to changes in observable characteristics, the second term reflects the change in the gap that is due to changes in the returns to the observable characteristics and the third term captures the change in the residual component over time. JMP (1993) deconstruct the change in the residual component into the change due to unobserved characteristics and the change attributable to unobserved prices. To show this, we can write our wage equation for natives in 2006 as:

where is a standardised residual with mean zero and variance one. In a similar fashion, we can write our wage equation for immigrants in 2006 as:

where

Using (5) and (6) we can re-write the change in the immigrant\native wage gap between 2006 and 2009 as:

where the third term in (7) captures the change due to unobserved characteristics and the fourth term reflects the change attributable to unobserved prices.

The decomposition allows us to separate out the impact of individual characteristics on the change in the wage gap over the period. However, there is an identification problem associated with separating out the effects of individual variables within the decomposition arising from the use of dummy variables where the number of categorical dummies exceeds one (Oaxaca and Ransom, 1999); essentially the change in wages attributable to differences in these types of variables may not be invariant to the choice of reference group. In such circumstances, we follow Gardeazabal and Ugidos (2004) and estimate the decompositions imposing a normalising restriction that the coefficients must sum to zero. The implementation of this restriction leaves the other coefficients unaffected.

4.2 Results

From Table 1, we can see that the raw immigrant\native wage gap was around 10 per cent in 2006 and the gap increased to 29 per cent in 2009. Here, we explore the drivers of the wage disadvantage of immigrants relative to natives over time. The decomposition outlined above indicates that changes in the wage gap over time depend on differences in the endowments of wage determining characteristics of natives and immigrants, variation in the average returns to these characteristics and a residual component. We begin by examining changes in characteristics of natives and immigrants over time and also differences in the returns to these characteristics. We then use the JMP decomposition to more formally explore the importance of changes in the composition and returns to specific attributes in explaining the raw immigrant\native wage gap.

Differences in Characteristics

Table 2 provides some basic descriptive statistics for natives and immigrants for 2006 and 2009. In 2006, relative to natives, immigrants are younger and have less tenure but they also have a higher level of educational attainment. There is also a higher share of immigrants working in the education sector and a smaller proportion working in public administration and defence, as compared to natives. Looking at the change in immigrant characteristics over the period, we can see that a lower proportion of immigrants in 2009 have degrees, that they are slightly younger and have somewhat lower tenure. Unfortunately, we do not have information on the year in which immigrants arrived in Ireland. To the extent that tenure and length of time in the country are correlated, the fact that immigrants in 2009 have, on average, lower tenure than in 2006 may indicate that those who lost their jobs were those who had been in Ireland for a longer period of time. This would tend to support the notion that foreign nationals with superior characteristics have lost their jobs and/or left the country. In addition, a lower proportion of immigrants have degrees and a smaller share work in the education sector. In terms of the change in native characteristics, the table shows that in 2009 a higher share of native workers are degree educated although they are, on average, slightly younger and have lower tenure than native workers in 2006. The big increase in the share of natives with degrees relative to the modest declines in average tenure and age may indicate that native workers with lower observable endowments have lost their jobs. This provides some preliminary evidence for the view that the observed rise in the unadjusted native\immigrant wage gap may be driven by compositional changes.

Table 2: Descriptive Statistics for Natives and Migrants in 2006 and 2009

* Note: Firm Size: 1: 1-9 employees, 2: 10-49 employees, 3: 50-249 employees, 4: 250-499 employees, 5: 500-999 employees, 6: 1000+ employees

Differences in Returns

In addition to compositional impacts, average hourly earnings are also expected to change due to variations in the average returns to these characteristics. To investigate this, we estimate OLS log hourly wage models for 2006 and 2009 and the results are presented in Table 3. The models include interaction terms to test for significant differences in the coefficients over time. In addition, we include a dummy variable to indicate those employees in the sample who are immigrants[footnoteRef:7] as well as a number of controls for human capital, job characteristics and industry characteristics. The results for both 2006 and 2009 indicate that immigrants earn significantly less than comparable natives. The earnings disadvantage is estimated to be 14.7 per cent in 2006. The results for 2009 indicate that although immigrants continue to earn less than comparable natives the gap has narrowed somewhat and the average earnings disadvantage is lower at 13.1 per cent. This fall in the immigrant\native wage gap could be evidence of an integration effect. The results also show a rise in the returns to having a degree and that the gender pay gap closed somewhat over the period. In terms of age groups, the estimates that all age groups earn less than those aged 40 to 49 (the reference group); however younger workers aged between 25 and 39 earned significantly less in 2009 than in 2006. There was also an increase in the return to working for a professional body. Finally, workers in industry and public administration and defence earned significantly more in 2009 than in 2006 while workers in construction and education earned significantly less. [7: The immigrant dummy variable is equal to one for all employees whose response to a question on nationality was anything other than ‘Irish’. Those who do not report their nationality are excluded from our sample.]

The wage disadvantage for immigrants is likely to hide differences across different types of immigrants. To explore this, we run the OLS wage models including separate dummies indicating the region the immigrant is from. The results are shown in Table 4. The estimates indicate that earnings for migrants from the UK are not significantly different to comparable natives in either year. Employees from the NMS experience the highest wage disadvantage and while there is no significant change in the earnings gap for EU-10 immigrants there is a fall in the earnings gap for EU-2 immigrants over the time period (significant at 10 per cent level). The results also show a large fall in the wage gap for immigrants from EU-15 (excluding UK and Ireland) countries. Finally, there is an increase in the wage disadvantage for immigrants from Non-EU English speaking countries while the wage gap for employees from Non-EU non-English speaking countries fell slightly.

Separate wage models were also run for employees in the private and public sector.[footnoteRef:8] The results indicate that there was no significant change in the wage disadvantage for immigrants in the private sector; immigrants in the private sector earned 15 per cent and 13.8 per cent less than comparable natives in 2006 and 2009 respectively. However, there was a significant fall in the wage disadvantage for immigrants in the public sector where the estimate falls from 15 per cent in 2006 to 4.8 per cent in 2009. [8: The results are available from the authors on request. The models include the same controls as those in Table 3.]

Table 3: OLS Wage Models

Note: *** p<0.01, ** p<0.05, * p<0.1

Table 4: OLS Wage Models with Immigrant Variable Broken Out

Note: *** p<0.01, ** p<0.05, * p<0.1. The same regressors as those in Table 2 are also included in the models presented in Table 3.

Decomposition Analysis

The descriptive data indicates that immigrant earnings fell by over 8 per cent between 2006 and 2009 while the earnings of natives increased by over 7 per cent the same period. The descriptive statistics also show that wage falls were exclusive to non-EU immigrants. These wage movements translated into a rise in the unadjusted\raw immigrant wage penalty from 10 per cent in 2006 to 29 per cent in 2009. However, care must be taken when interpreting such data as the observed increase in the immigrant penalty can be driven be either a higher concentration of lower skilled workers within the immigrant population in Ireland (an endowment effect) and / or a deterioration in rates of return to wage determining characteristics among immigrants relative to their native counterparts (a coefficient effect).

Separate Oaxaca decompositions[footnoteRef:9] for the raw immigrant wage gap in 2006 and 2009 indicate that in 2006 around -50 per cent of the immigrant wage gap was attributable to changes in characteristics, indicating that immigrants had superior endowments to natives, while 150 per cent of the immigrant wage gap was attributable to coefficient effects. However, by 2009, approximately 50 per cent of the immigrant wage disadvantage was related to differences in the characteristic make up between immigrants and natives (and only 33 per cent was attributable to coefficient effects), suggesting that substantial changes took place in the characteristic make up of either the immigrant or native employee populations during the period. [9: The results of the Oaxaca decompositions are available from the authors.]

In order to determine the source of the deterioration in the unadjusted immigrant\native wage gap over time we estimate a John Murphy Peirce (JMP) decomposition the results of which are presented in Table 5. Based on the model coefficients, the JMP approach predicts that the raw immigrant wage differential will have widened by 17 per cent points between 2006 and 2009[footnoteRef:10] with most of this change attributable to observable effects; the change in the predicted gap or the change due to observable characteristics and returns to these characteristics was 19 per cent, while the change in the residual gap was -2 per cent. Specifically, just under 30 per cent of the growth in the predicted differential (19 per cent) was due to a fall in the share of graduates within the immigrant population while a further 26 per cent of the decline a fall in the number of immigrants employed within relative well paid public sector occupations (Table 5). [10: This aligns closely with the descriptive results.]

Table 5: Juhn-Murphy Pierce Decomposition of the Change in the Immigrant Wage Gap Between 2006 and 2009

In order to pinpoint the source of the endowment effects driving the rise in the unadjusted immigrant wage disadvantage, we estimate JMP models for both the public and private sectors and for each immigrant grouping.[footnoteRef:11] While the predicted raw immigrant wage penalty increased in both the public and private sectors, there were substantial differences in the scale of the effect across both sectors. Within the private sector, the predicted raw immigrant wage penalty increased by 2.9 percentage points between 2006 and 2009, and again most of this is attributable to changes in observable effects. The change in the predicted gap is explained by endowment and coefficient impacts in almost equal measures. With respect to private sector endowment impacts, these were dominated by falls in the employment share of graduates which resulted in a rise in the immigrant pay differential of 2.1 percentage points. Within the public sector, the predicted raw immigrant wage gap increased by a remarkable 20 percentage points over the period, with all of the movement explained by changes in the composition of the immigrant workforce. The bulk of the change in the predicted public sector immigrant pay disadvantage can be explained by declines in the following among the population of publically employed immigrants (a) the level of employment within the Education sector (38%), (b) the number of graduates (27%), and (c) the average level of worker tenure (19%). [11: Detailed results available from the authors. ]

Finally, with respect to the relative effects across immigrant groupings, in line with the descriptive data, we found that immigrants from non-EU non-English speaking countries experienced the most substantial deterioration over the period. Between 2006 and 2009 the predicted raw pay differential of immigrants from non-EU non-English speaking countries employed within all sectors of the economy increased by 36 percentage points relative to their native counterparts. All of the decline in the wage position of this immigrant group was again attributable to composition influences related to a falls in the number of graduates and workers employed within the Education sector; in addition, a fall in average tenure levels and a drop in the number of person employed in large firms also contributed to a fall in relative wages.

It is not possible to discern from the data the exact reasons underlying the changes observed within the immigrant wage penalty. Nevertheless, by 2009 the impacts of the fiscal crises were being fully felt with substantial pressure being placed on public sector organisations to cut costs. The industrial relations framework in Ireland precluded the widespread use of redundancy as a cost control mechanism; however, it is likely that departments sought to lower wage costs through the non-renewal of fixed term or temporary contracts and, presumably, such policies disproportionately effecting graduate immigrants from non-EU non-English speaking countries.

To summarise, the raw data shows a large increase in the immigrant wage penalty over the crisis period. However, when we control for relevant characteristics, the OLS results indicate that the immigrant wage penalty fell slightly over the period. This could point to an integration effect in the labour market. Unfortunately, our dataset does not contain information on date of arrival; however, the data show that average tenure has fallen among immigrants, so to the extent that tenure and length of stay are correlated, the fall in the wage penalty is unlikely to reflect an integration effect. Our results also show differences in the immigrant penalty across different migrant groups. For example, immigrants from the NMS experience the largest overall pay penalty - the gap for EU-10 immigrants remained unchanged between 2006 and 2009 while the gap for EU-2 immigrants narrowed over the period. In addition, we find a big fall in the penalty for EU-15 (excluding UK and Ireland) immigrants. The decomposition results show that compositional changes are driving the change in the raw immigrant wage penalty. In particular, a fall in the share of immigrants with degrees and a fall in the share of immigrants in the relatively well-paid public sector occupations explain a substantial part of the change in the wage gap.

5. Conclusions

Ireland’s economic and migratory experiences over the last fifteen to twenty years have been dramatic. The economic boom provided a remarkable period of growth which led to a reversal of Ireland’s traditional pattern of outward migration. The rate of inflow was large and produced something of a transformation in the make-up of the Irish population. For example, in the four year period 2002 to 2006, the proportion of Ireland’s population that was non-Irish grew from 7 percent to 11 percent (Barrett, 2010). The economic collapse has led to the resumption of net out-flows, with both Irish people and immigrants contributing.

As a result of these movements, Ireland provided an interesting new case-study for migration research. Many of the findings on Ireland’s immigrants which were produced during the boom mirrored results that had been found elsewhere. For example, immigrants earned less compared to natives and were found to have had a positive effect on economic growth. Other research findings showed differences between Ireland and elsewhere. An example here was the finding that immigrants were less likely to receive welfare supports.

The economic collapse was more severe in Ireland relative to elsewhere. Many questions arose on the impacts of the collapse, including the impacts of immigrants and their subsequent reactions. Earlier research showed that immigrant employment contracted sharply over the recession, thereby suggesting reduced demand for immigrant labour. In this chapter, we have expanded the analysis of the impacts of the recession on immigrants by asking if their earnings also fell, relative to natives. Although the raw data show a widening of the immigrant/native pay gap, a decomposition analysis shows that most of this was generated by the changing composition of the immigrants who were employed.

Both the immigrant population and the native population in Ireland are now reacting to the downturn, to an extent, through out-migration. Net outflows from Ireland resumed in 2010 and in the three year period ending April 2012, it is estimated (by the Central Statistics Office) that almost 90,000 people (net) have left Ireland. This represents about 2 percent of the population of 2010. As discussed above, the rate of net outflow is higher for EU NMS immigrants in particular. For this reason, Ireland seems to be benefiting from a relatively mobile labour force which flowed in during the boom and which is flowing out during the downturn. This makes the Irish situation somewhat different to elsewhere where less subsequent mobility of immigrant populations is observed. The mobility observed in Ireland may have been a function of the fact that many immigrants had arrived in the years leading up to the downturn and so may not have had time to become rooted in Ireland. Alternatively, the fact that the Irish government restricted welfare to immigrants may have reduced the incentive to stay (Barrett, forthcoming 2013). Either way, flows to and from Ireland will continue to provide interesting insights into migratory mechanisms and their effects.

References

Barrett, A. (2010). “EU Enlargement and Ireland’s Labour Market”, (2009), chapter 6 in M. Kahanec and K. F. Zimmermann (eds.), EU Labor Markets after Post-Enlargement Migration, Berlin: Springer.

Barrett, A., A. Bergin and D. Duffy (2006). “The Labour Market Characteristics and Labour Market Impacts of Immigrants in Ireland”, Economic and Social Review Vol. 37 No. 1.

Barrett, A. and D. Duffy (2008). “Are Ireland’s Immigrants Integrating into its Labour Market?”, (2008), International Migration Review Vol. 42 No. 3.

Barrett, A., C. Joyce and B. Maïtre (forthcoming 2013) “Immigrants and Welfare Receipt in Ireland”, International Journal of Manpower, Vol. 34 No. 1/2

Barrett, A. and E. Kelly (2012). “The Impact of Ireland’s Recession on the Labour Market Outcomes of its Immigrants”, European Journal of Population, Vol. 28, No. 1, pp. 99-111.

Barrett, A. and Y. McCarthy (2007). “Immigrants in a Booming Economy: Analysing their Earnings and Welfare Dependence”, LABOUR: Review of Labour Economics and Industrial Relations Vol. 21 No. 4-5.

Barrett, A. and Y. McCarthy (2008). “Immigrants and Welfare Programmes: Exploring the Interactions between Immigrant Characteristics, Immigrant Welfare Dependence and Welfare Policy”, Oxford Review of Economic Policy Vol. 24 No. 3

Barrett, A., McGuinness, S., and M. O’Brien (2012). “The Immigrant Earnings Disadvantage across the Earnings and Skills Distributions: The Case of Immigrants for the EU's New Member States", British Journal of Industrial Relations , Vol. 50, Issue 3, September 2012, pp. 457-481.

Gardeazabal, J. and A. Ugidos (2004). “More on Identification in Detailed Wage Decompositions.” The Review of Economics and Statistics, Vol. 86, No. 4, pp. 1034-6.

Juhn, C., K.M. Murphy and B. Pierce (1993). “Wage Inequality and the Rise in Returns to Skill”, Journal of Political Economy, Vol. 101, No. 3, pp. 410-442.

Oaxaxa, R. and M. Ransom (1999). “Identification in Detailed Wage Decompositions”, The Review of Economics and Statistics, Vol. 81, No. 1, pp. 154-157.

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 -23 -41.9 -43.9 -22.9 -2 7.4 -0.4 -4.7 -1.9000000000000001 8 19.2 17.399999999999999 17.3 26 32.800000000000004 41.3 30.7 32 55.1 71.8 104.8 64.3 1.6 -27.5 -27.4 -34.4 Irish Q3 07 Q3 08 Q3 09 Q3 10 Q3 11 Q3 12 2.1642124051762582E-2 -1.7580257698187302E-2 -7.9304212515282912E-2 -2.5653407376109178E-2 -2.8497088340973857E-2 2.1680908047442085E-3 All non-Irish Q3 07 Q3 08 Q3 09 Q3 10 Q3 11 Q3 12 0.18347338935574234 -2.9585798816568051E-3 -0.11928783382789306 -8.4231805929919232E-2 2.0971302428256091E-2 -2.8468468468468386E-2 EU NMS Q3 07 Q3 08 Q3 09 Q3 10 Q3 11 Q3 12 0.37846655791190903 2.9585798816568051E-3 -0.18466076696165187 -0.1063675832127351 2.5910931174089089E-2 -3.9463299131807421E-3 GDP 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 9.3225403379149779E-2 0.11495552353028893 8.7915761323750646E-2 0.11051364776975472 0.1073922065944201 5.2961213968634839E-2 5.637200 6201157981E-2 3.8845419490097781E-2 4.3627426248531602E-2 5.8765435509268341E-2 5.4042594827305321E-2 5.4452627018999014E-2 -2.1087042003885319E-2 -5.4563658600925831E-2 -7.6603570182948236E-3 1.4307897780646315E-2 GNP 8.9566740765822758E-2 0.10052559754631087 7.8194251758726338E-2 8.8840810749260229E-2 0.11059170483779894 3.1130822904601875E-2 2.0273417166653082E-2 4.8745558210932377E-2 3.9805750667838841E-2 5.7874303349667172E-2 6.4096396759877328E-2 4.1583725254292901E-2 -1.7725134408602222E-2 -8.0732061917386702E-2 9.4117282227804168E-3 -2.4723122532680001E-2

19

15000

20000

25000

30000

35000

40000

45000

1

9

9

7

Q

1

1

9

9

7

Q

3

1

9

9

8

Q

1

1

9

9

8

Q

3

1

9

9

9

Q

1

1

9

9

9

Q

3

2

0

0

0

Q

1

2

0

0

0

Q

3

2

0

0

1

Q

1

2

0

0

1

Q

3

2

0

0

2

Q

1

2

0

0

2

Q

3

2

0

0

3

Q

1

2

0

0

3

Q

3

2

0

0

4

Q

1

2

0

0

4

Q

3

2

0

0

5

Q

1

2

0

0

5

Q

3

2

0

0

6

Q

1

2

0

0

6

Q

3

2

0

0

7

Q

1

2

0

0

7

Q

3

2

0

0

8

Q

1

2

0

0

8

Q

3

2

0

0

9

Q

1

2

0

0

9

Q

3

2

0

1

0

Q

1

2

0

1

0

Q

3

2

0

1

1

Q

1

2

0

1

1

Q

3

2

0

1

2

Q

1

E

u

r

o

M

i

l

l

i

o

n

s

GDP

GNP

NES 2006

NES 2009