Abstract
The Selected Issues paper analyzes how fast Ireland will grow in the future. The approach of this paper is to consider the catch-up in labor utilization productivity and use independent demographic projections and other considerations to make reasonable assumptions about labor productivity and utilization growth in the future. It uses a simple growth-accounting framework, and discusses the trends in labor utilization and productivity per hour in the past. The paper also describes the spectacular boom of the Irish housing market and its key drivers from an international perspective.
I. Potential Growth After the Boom1
A. Introduction
1. Ireland experienced a period of unprecedented growth in the 1990s, catching up and even overtaking other industrial countries on a per capita income basis (see Table 1). Between 1993 and 2003, real income per capita rose by a cumulative 71 percent (94 percent), when GNP (GDP) is used as a measure of aggregate income. As a result of this impressive growth, Ireland’s income levels mostly converged to those of other industrial countries on a per capita basis. In 2003, real GNP per capita in Ireland was about 99 percent of the EU average, while real GDP per capita was almost 20 percent above it.
Average Annual Growth Rate of GDP per Capita 1/
In 1995 PPPs, US dollars.
Average Annual Growth Rate of GDP per Capita 1/
| 1961-70 | 1971-80 | 1981-90 | 1991-2000 | |
|---|---|---|---|---|
| Australia | 3.4 | 1.5 | 1.5 | 2.3 |
| Austria | 4.1 | 3.5 | 2.1 | 1.9 |
| Belgium | 4.3 | 3.2 | 1.9 | 1.9 |
| Canda | 3.3 | 2.7 | 1.5 | 1.9 |
| Denmark | 3.9 | 1.6 | 1.6 | 1.9 |
| Finland | 4.4 | 3.5 | 2.6 | 1.6 |
| France | 4.5 | 2.7 | 2.0 | 1.5 |
| Germany | 3.5 | 2.6 | 2.0 | 0.5 |
| Iceland | 3.2 | 5.2 | 1.7 | 1.7 |
| Ireland | 3.8 | 3.2 | 3.3 | 6.4 |
| Ireland (GNP) | 3.6 | 2.2 | 2.8 | 5.8 |
| Italy | 5.0 | 3.2 | 2.2 | 1.4 |
| Japan | 9.0 | 3.3 | 3.4 | 1.2 |
| Netherlands | 3.7 | 2.1 | 1.7 | 2.3 |
| New Zealand | 1.8 | 0.7 | 1.2 | 1.5 |
| Norway | 3.4 | 4.3 | 2.2 | 3.1 |
| Spain | 6.3 | 2.5 | 2.6 | 2.4 |
| Sweden | 3.9 | 1.6 | 1.9 | 1.7 |
| UK | 2.3 | 1.9 | 2.4 | 2.2 |
| US | 2.9 | 2.2 | 2.2 | 2.3 |
In 1995 PPPs, US dollars.
Average Annual Growth Rate of GDP per Capita 1/
| 1961-70 | 1971-80 | 1981-90 | 1991-2000 | |
|---|---|---|---|---|
| Australia | 3.4 | 1.5 | 1.5 | 2.3 |
| Austria | 4.1 | 3.5 | 2.1 | 1.9 |
| Belgium | 4.3 | 3.2 | 1.9 | 1.9 |
| Canda | 3.3 | 2.7 | 1.5 | 1.9 |
| Denmark | 3.9 | 1.6 | 1.6 | 1.9 |
| Finland | 4.4 | 3.5 | 2.6 | 1.6 |
| France | 4.5 | 2.7 | 2.0 | 1.5 |
| Germany | 3.5 | 2.6 | 2.0 | 0.5 |
| Iceland | 3.2 | 5.2 | 1.7 | 1.7 |
| Ireland | 3.8 | 3.2 | 3.3 | 6.4 |
| Ireland (GNP) | 3.6 | 2.2 | 2.8 | 5.8 |
| Italy | 5.0 | 3.2 | 2.2 | 1.4 |
| Japan | 9.0 | 3.3 | 3.4 | 1.2 |
| Netherlands | 3.7 | 2.1 | 1.7 | 2.3 |
| New Zealand | 1.8 | 0.7 | 1.2 | 1.5 |
| Norway | 3.4 | 4.3 | 2.2 | 3.1 |
| Spain | 6.3 | 2.5 | 2.6 | 2.4 |
| Sweden | 3.9 | 1.6 | 1.9 | 1.7 |
| UK | 2.3 | 1.9 | 2.4 | 2.2 |
| US | 2.9 | 2.2 | 2.2 | 2.3 |
In 1995 PPPs, US dollars.
2. The main question addressed in this paper is how fast will Ireland grow in the future. With the convergence process mostly complete, this is a difficult question to answer, as the past may not be the best guide to the future. Nevertheless, the approach of this paper is to consider the catch up in labor productivity and utilization and use independent demographic projections and other considerations in order to make reasonable assumptions about labor productivity and utilization growth in the future.
3. The rest of the paper is organized as follows. Using a simple growth-accounting framework, Section B discusses the trends in labor utilization and productivity per hour in the past. Section C presents our baseline projections for labor productivity growth and the components of labor utilization growth in the future. Section D concludes.
B. Decomposing Past Output Growth
4. Ireland’s impressive economic performance over the past decade raises the question of whether such a boom can be repeated in the future. Although productivity growth (measured as output per hour worked) was strong throughout the period peaking at 4.2 percent a year during the mid-1990s—well above the productivity growth in most industrial countries—what set Ireland apart was the extraordinary increase in labor utilization during those years. Therefore, whether Ireland could replicate the remarkable boom of the 1990s largely depends on future trends in labor utilization. This section introduces a simple growth accounting framework to examine the sources of Ireland’s growth in the late 1990s.
Decomposition of GNP Growth
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: Central Statistics Office; and OECD.Decomposition of GNP Growth
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: Central Statistics Office; and OECD.Decomposition of GNP Growth
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: Central Statistics Office; and OECD.Growth-accounting framework
5. Output growth can be decomposed into several components. Using a simple identity, one can express output per capita (1) as the product of labor productivity per hour (2), average hours worked (3), the employment rate (4), the participation rate (5), and the inverse of the dependency ratio (6):
where Y is output, N is total population, Wp is working-age population, Lf is labor force, L is total employment, and h is average hours worked per employee. Therefore, per capita output growth is equal to the sum of labor productivity growth and the growth in the above four components of labor utilization (see (3) − (6)). By the same token, output growth can be calculated as the sum of labor productivity growth and the five components of labor utilization growth (average hours, employment rate, participation rate, inverse of dependency ratio, and population).
Past growth in labor utilization
6. Similar to other industrial countries, average hours worked in Ireland have fallen steadily since the mid-1970s, which can be attributed to several factors (see Figure 1). First, the rise in female participation, especially during the mid-1990s, led to a shift from full-time to part-time employment and was associated with fewer hours worked. Second, faster growth in services relative to the industrial and agricultural sectors has also contributed to the decline, given the greater use of part-time employment and fewer average hours worked in the service sector. Third, strong preference for leisure and the Working Time Act may help explain the decline in the average hours worked by full-time workers.2
Ireland: Factors Explaining Average Hours
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: CSO, Eurostat, and OECD.1/ Share of part-time in employment in percent.2/ Average hours worked per employee.3/ Share of part-time employment in percent.4/ Services excluding public administration.Ireland: Factors Explaining Average Hours
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: CSO, Eurostat, and OECD.1/ Share of part-time in employment in percent.2/ Average hours worked per employee.3/ Share of part-time employment in percent.4/ Services excluding public administration.Ireland: Factors Explaining Average Hours
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: CSO, Eurostat, and OECD.1/ Share of part-time in employment in percent.2/ Average hours worked per employee.3/ Share of part-time employment in percent.4/ Services excluding public administration.7. After sluggish performance during most of the 1980s, the employment rate started rising sharply in the early 1990s. While a number of factors contributed to the lack of job creation during the 1980s, poor demand management policies are often cited as the primary reason, resulting in high taxes and interest rates.3 Faced with unemployment rates of over 16 percent, the social partners—employees, employers and the government—decided to adopt a cooperative approach to wage setting in the late 1980s, based on trading wage moderation and industrial peace for tax cuts and social welfare improvements. While not the only factor, the social partnership contributed significantly to the increase in the employment rate since the early 1990s, which averaged 2.1 percent per annum during the second half of the 1990s.
Employment Rate
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Source: Central Statistics Office.Employment Rate
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Source: Central Statistics Office.Employment Rate
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Source: Central Statistics Office.8. The participation rate also picked up, mainly reflecting higher female participation. Following a steady decline from the mid 1960s to the late 1980s, the participation rate started increasing in the early 1990s. This increase was mainly driven by higher female participation, which, in turn, reflected better job opportunities as a result of the improving economy. Male participation also rose in the 1990s, albeit at a much slower pace.
Participation Rates
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Participation Rates
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Participation Rates
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Female Participation
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Female Participation
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Female Participation
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
9. Providing an additional boost to economic growth, the inverse dependency ratio increased substantially, reflecting favorable demographic factors. In contrast to other European countries, fertility rates in Ireland were very high in the 1960s and 1970s. As a result, the working-age population increased significantly in the 1980s and 1990s, prompting a steady rise in the inverse dependency ratio.
Inverse Dependency Ratio 1/
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: Central Statistics Office; and staff calculations.1/ Working -age population over total population.Inverse Dependency Ratio 1/
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: Central Statistics Office; and staff calculations.1/ Working -age population over total population.Inverse Dependency Ratio 1/
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: Central Statistics Office; and staff calculations.1/ Working -age population over total population.
Fertility Rates
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Source: Eurostat.Fertility Rates
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Source: Eurostat.Fertility Rates
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Source: Eurostat.10. Population growth also supported the increase in output, helped by a reversal of migration flows. In addition to Ireland’s higher fertility rates, the turnaround in migration flows also played an important role in boosting population growth, starting in the mid 1990s. The net inflow of migrants to Ireland between 1996 and 2003 was close to 0.2 million (compared to a total population of about 4 million in 2003).
Past growth in labor productivity per hour
11. In Ireland, labor productivity is notoriously difficult to measure, given the distortions to sectoral output caused by the operations of multinational companies. It is a well known fact that the level and growth in labor productivity in Ireland has been higher in industries dominated by foreign companies. And while this superior performance during the 1990s also reflected true productivity improvements (as a result of the surge in FDI and the associated technology transfer), a nontrivial part of it was due to monopoly (or patent-related) profits of multinational companies that were booked in Ireland, given its low corporation tax rate regime. In other words, as Honohan and Walsh (2002) point out, “… in many cases, the huge profits recorded by the Irish affiliates [had] very little to do with the manufacturing activities being conducted in Ireland”, but rather emerged as a consequence of transfer pricing. In this context, it is not surprising that capturing true productivity growth in Ireland has been a major challenge. In this section, we consider three measures of aggregate labor productivity—GDP per hour, GNP per hour, and adjusted GDP per hour. The unadjusted GDP measure includes the profits of foreign-owned firms operating in Ireland, while the GNP measure excludes them. The construction of the adjusted GDP measure is described below.
12. How can we make an adjustment for the impact of the multinational sector on measured productivity? To start, we compare Ireland’s labor productivity levels in four industries dominated by multinationals—chemicals, printing and publishing, office machinery, and electronic valves and tubes—to the average productivity levels of the same industries in the EU.4 As expected, productivity levels in these industries—particularly in chemicals and printing and publishing—rose much faster in the 1990s than in any other country. For example, measured labor productivity in Ireland’s chemical industry, which exceeded the EU average by about 150 percent in 1990, shot up to over 350 percent in 2001! The corresponding figures for the printing and publishing industry are similar. Then, to adjust for labor productivity distortions, we assume that in the absence of the multinational sector in the second part of the 1990s, Ireland’s productivity per hour in these two industries would have grown at the highest rate observed across all other countries in the sample.
13. Labor productivity growth depends on which measure of productivity is used to construct it. Prior to the mid-1990s, the difference between the alternative measures does not, on average, exceed 0.5 percentage points. However, in the period from 1995 to 1999, the gap between the growth rates of GNP per hour (4.2 percent) and adjusted GDP per hour (4.2 percent) and the growth rate of GDP per hour (5.5) is particularly stark. In other words, the GNP and adjusted GDP measures indicate a substantially lower productivity growth in the second half of the 1990s than the GDP measure.
Growth in Labor Productivity per Hour
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Growth in Labor Productivity per Hour
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Growth in Labor Productivity per Hour
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
14. In the remaining part of the paper, we focus on GNP per hour as the most appropriate measure of productivity, given the problems associated with the other two measures. The use of unadjusted GDP is clearly inappropriate as already suggested in para 10. As far as adjusted GDP is concerned, there are a number of other assumptions that one could make about counterfactual developments in the industries dominated by multinational companies. Unfortunately, the adjusted GDP measure is not robust to the specific assumption of what Irish productivity growth would have been in the multinational-dominated industries if the foreign companies had not entered the Irish market, as the magnitude of the adjustment could change substantially under alternative assumptions. Hence, we are left with GNP per hour as a more suitable measure of productivity than its two alternatives.
Past growth in output
15. The decomposition of GNP growth illustrates that higher growth in both labor productivity and labor utilization explain Ireland’s boom in the 1990s. Putting together the components of output growth discussed in this section shows the relative contributions to GNP growth of the variables discussed in this section. During the second half of the 1990s, productivity growth accounted for slightly higher than ½ of the overall growth rate, while the sharp rise in employment and participation rates was mainly responsible for the rest.
Decomposition of GNP Growth
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: Central Statistics Office; and OECD.Decomposition of GNP Growth
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: Central Statistics Office; and OECD.Decomposition of GNP Growth
(In percent)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Sources: Central Statistics Office; and OECD.C. Aggregating Future Output Growth
Future growth in labor productivity per hour
16. Using the appropriate measure of output, Ireland’s hourly productivity level still has room to catch up with those of the leaders. In 2003, GNP per hour in Ireland was about 90 percent of the corresponding level in the US.5 Therefore, it is reasonable to allow for a further convergence in the Irish productivity level and assume that the country’s productivity growth would continue to exceed that of the US during the next decade.
17. To allow for further convergence in labor productivity levels, we assume a prospective growth rate in trend productivity of 3 percent. This is based on the expectation that prospective hourly labor productivity growth in the US is 2 percent (which is consistent with its recent performance) and that Ireland continues to close the gap in productivity levels at a rate of 1 percent per annum. In the context of its historical performance, the assumed GNP per hour growth of 3 percent growth is about ½ percent lower than in 1990-2003 and 1½ percent lower than during the boom period, 1995–2000.
Future growth in labor utilization
18. Going forward, average hours worked are expected to be broadly stable. Most of the factors contributing to the reduction in average hours worked in the 1990s appear to have run their course. In particular, the increase in part-time employment seems to have petered out. Given the limited scope for further increases in female participation, we expect part-time employment to remain broadly stable as a proportion of total employment. Average full-time hours are also assumed to stabilize at around 41 hours a week, barring new labor legislation and reflecting weaker income effects. The dynamics of hours worked in recent years seem consistent with the assumption of no trend in average hours worked in the future. However, a further compositional shift in employment towards services, leading to an additional decline in average hours worked, may be a downside risk to our baseline assumption.
Average Weekly Hours
(Full-time)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Source: Eurostat.Average Weekly Hours
(Full-time)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Source: Eurostat.Average Weekly Hours
(Full-time)
Citation: IMF Staff Country Reports 2004, 349; 10.5089/9781451818802.002.A001
Source: Eurostat.19. The employment rate is assumed to increase modestly. Future trends in the employment rate are determined by prospective developments in labor supply and employment. On labor supply, our projections are based on the latest estimates made by the Economic and Social Research Institute (ESRI), which show a strong growth of about 2 percent per annum until 2010 and a somewhat slower growth of 0.9 percent per annum afterwards.6 On employment, we also use the ESRI medium-term projections, which indicate an average growth rate of over 2 percent per annum until 2010 and a more modest growth of 1.1 percent per annum in later years. Therefore, we assume employment growth of 1.5 percent on average up to 2015, consistent with a NAIRU estimate of 4 percent, while labor force average growth is projected at 1.45 percent. Hence, our assumptions imply an annual growth in the employment rate of 0.05 percent per annum—substantially lower than that during the Celtic Tiger era.
20. The participation rate is expected to rise further but at a slower rate, consistent with falling fertility rates and declining immigration flows. Prospective growth in the participation rate depends on future growth in labor supply and working-age population. On labor supply, we use the assumptions already described in para. 19. On working-age population, we use the ESRI’s assumptions of an average growth of 0.8 percent in the period up to 2015, consistent with slowing fertility rates and stabilizing net migration to 10,000 persons by 2015. Consequently, average growth in the participation rate is projected to be about 0.65 percent per annum until 2015.
21. The inverse dependency ratio is assumed to decline slightly. Growth in the inverse dependency ratio reflects the differential growth rates of working-age and total population. As discussed in para. 20, the prospective growth of working-age population is taken to be 0.8 percent per annum. Therefore, we project the inverse dependency ratio to decrease by 0.35 percent per annum, consistent with a population growth of about 1.15 percent.
22. Population growth is projected to slow down. While estimates of future population growth differ across institutions, its slowdown is undisputable (see Table 2). Over the medium-term, population growth is expected to slow down, reflecting a decline in fertility rates and a stabilization in net migration. As already mentioned in para. 21, we assume that Ireland’s population grows by an average of 1.15 percent per annum until 2015 in line with the ESRI’s estimates.
Population Estimates
(Average annual growth rate, in percent)
Central Statistics Office (2001), M1F2 scenario.
Baseline scenario, 1999.
Database for United Nations (2003).
Future growth in potential output
23. Aggregating its components, we project potential GNP growth of 4½ percent in the medium run. The baseline projections for labor productivity per hour and the five components of labor utilization are summarized in Table 3. As already suggested in the previous section, labor supply growth was the most important factor explaining the pick up in output growth during the mid-1990s. In the absence of a similar increase in labor supply over the next decade the projected trend growth is 4½ percent.
Decomposition of Potential Growth
(In percent)
Decomposition of Potential Growth
(In percent)
| Productivity | 3.00 |
| Average hours | 0.00 |
| Employment | 0.05 |
| Participation | 0.65 |
| Inverse dependency | −0.35 |
| Population | 1.15 |
| Potential output | 4.50 |
Decomposition of Potential Growth
(In percent)
| Productivity | 3.00 |
| Average hours | 0.00 |
| Employment | 0.05 |
| Participation | 0.65 |
| Inverse dependency | −0.35 |
| Population | 1.15 |
| Potential output | 4.50 |
D. Concluding Remarks
24. This paper presents a projection for Ireland’s potential growth over the next decade and explains its underlying assumptions. Using a simple growth accounting framework, we make reasonable assumptions about future growth in labor productivity, average hours worked, employment and participation rates, the inverse dependency ratio, and population. Adding up these components, we arrive at an estimate of potential output growth of 4½ percent over the medium run.
25. Nonetheless, there are significant, mostly downside risks to our baseline projection for potential growth. First, productivity growth may be lower than 3 percent, particularly if Ireland loses its attractiveness as a destination for FDI. Second, hours worked may continue to fall, reflecting stronger-than-expected income effects and preference for leisure. Third, migration flows may be lower than expected, especially if high property prices act as a constraining force on labor supply. Finally, female participation may reverse its historical trend and decline, particularly if childcare costs continue to rise relative to wages.
References
Bergin, A., Cullen. J., Duffy, D., Fitz Gerald, J., Kearny, I., and McCoy, D., 2003, Medium-Term review: 2003–2010, (Dublin: Economic and Social Research Institute).
Central Statistics Office, 2001, Regional Population Projections 2001-2031, (Dublin: Central Statistics Office).
Honohan., P., and B. Walsh, 2002, “Catching up with the leaders: the Irish hare,” Brookings Papers on Economic Activity: 1, pp. 1–77.
EPC-Economic Policy Committee, 2001, Budgetary Policy Challenges Posed by an Aging Population”, (Brussels: Economic Policy Committee). Available via the internet: http://europa.eu.int/comm/economy_finance/epc/documents/2003/awgeducation_en.pdf
European Commission, 2004, Illustrative Long-run Growth Projections for the EU15 and AC10 Countries up to 2050, (Brussels: European Commission).
OECD, 2003, Economic review of Ireland, (Paris: OECD).
O’Mahony and van Ark, 2003, Industry Labour Productivity Database.
United Nations, 2003, World Population Prospects: the 2002 Revision, (New York: United Nations).
Prepared by Petya Koeva and Marialuz Moreno Badia.
The Organization of Working Time Act was introduced in 1997 to provide for the implementation of Directive 93/104/EC of 23 November 1993 of the Council of the European Communities concerning certain aspects of the organization of working time. It established a maximum of 48 hour working week averaged over a reference period; a minimum daily rest period of 11 consecutive hours a day; a rest break where the working day is longer than four hours; a minimum rest period of one day a week; a statutory right to annual paid holiday of four weeks; and a maximum night working time of eight hours a night, on average.
We use the Industry Labour Productivity Database by O’Mahony and van Ark (2003), which contains data on labor productivity per hour in EU coun]tries and the US from 1979 to 2001. The database covers 57 industries, including services.
In the same year, Ireland’s GDP-based hourly productivity level surpassed that of the US.
