Ireland: Selected Issues and Statistical Appendix

This Selected Issues paper and Statistical Appendix on Ireland examines the productivity growth in Irish traded and nontraded goods, and provides some rough estimates of the sort of wage and inflation differentials that would be predicted by a Balassa–Samuelson framework under certain growth assumptions for the future. The paper provides a framework for judging what sort of wage growth and inflation could be sustained over the medium term without leading to a loss of competitiveness. The paper also examines traded and nontraded productivity in Ireland.

Abstract

This Selected Issues paper and Statistical Appendix on Ireland examines the productivity growth in Irish traded and nontraded goods, and provides some rough estimates of the sort of wage and inflation differentials that would be predicted by a Balassa–Samuelson framework under certain growth assumptions for the future. The paper provides a framework for judging what sort of wage growth and inflation could be sustained over the medium term without leading to a loss of competitiveness. The paper also examines traded and nontraded productivity in Ireland.

II. Potential Output Growth in Ireland16

A. Introduction

1. Since the beginning of the 1990s Ireland’s GDP has grown by a cumulative 62½ percent in real terms; the increase in GNP was about 55 percent over the same period. This growth has been achieved by sizable foreign direct investment flows (FDI), and a rapidly growing labor force, and has been supported by strong productivity growth. The unpredictable nature of foreign direct investment and the highly elastic labor supply make it extremely difficult to estimate, with a reasonable degree of accuracy, the potential output level in Ireland. Potential output can be thought of as “the maximum production without inflationary pressure—or more precisely—the point of balance between more output and greater stability” (Okun, 1970, pp. 132–33).

2. Given that monetary policy is decided at the euro-area level and the size, or the sign for that matter, of the Irish output gap will make little difference in the decision of the ECB to relax or tighten monetary policy it could be argued that information about the output gap may be irrelevant. However, there are at least two basic reasons why it may be important to obtain accurate estimates of potential output for the Irish economy. First, reliable information about potential output growth makes it possible to obtain better forecasts about the growth prospects in the medium term. Second, it allows policymakers to gauge the stance of fiscal policy.

3. The main focus of this paper is, therefore, to estimate the potential output for Ireland. A number of different measures of the output gap are compared but the preferred method is the production function approach that decomposes the potential into changes in the capital stock, labor, and productivity. A second objective is to evaluate the prospects for the medium term. The discussion is motivated by reference to the neoclassical growth model, which predicts that economies with lower capital per person tend to grow faster in per capita terms. In other words, it suggests that there will be convergence across economies. A corollary of this hypothesis is that growth will slow down in the medium term and that the potential of the Irish economy to carry on growing rapidly without significant wage and price inflation may diminish.

B. Comparing Different Estimates of Potential Output

4. A number of recent papers have reviewed extensively the different methods for estimating potential output, for example, Adams and Coe (1990), Bayoumi (1999), Canova (1998), European Commission (1999), Kenny (1996), Magnier (1996), and references therein. Most of these approaches are based on purely statistical techniques, while few others rely mostly on economic theory; consequently, the results from these different methods vary considerably although most yield identical sign for the output gap.

5. Three methods are used in this study; the HP filter, the production function approach and finally an unobserved component model. The estimation of the output gap using the latter two approaches are discussed in Section C and D. The most commonly used statistical method, the HP filter, optimally extracts a smooth, stochastic trend from the data that is uncorrelated with the cyclical component. The parameter λ determines how smooth the trend component is relative to the actual series. The essence of the production function approach is an explicit modeling of output in terms of underlying factor inputs. Thus it involves the modeling of a production function that links output to factor inputs and the determination of the levels of inputs. Finally, the unobserved component model uses observed times series, such as GDP and inflation, to draw conclusions about the unobserved potential output (Kuttner, 1994).

6. A number of different estimates of the output gap for Ireland are shown in Table 1. The output gaps (as a percent of potential) for three episodes during the 1990s are shown in Table 1: the peak of 1990, the following trough in 1993, and the most recent boom in 1998 (which is not yet considered a turning point).

Table 1.

Ireland: Output Gap Estimates

(Percent of potential)

article image

7. The top two rows show the growth rates of GDP and GNP. Given the large difference between GDP and GNP in Ireland, we provide estimates of potential output based on both series; the top panel of Figure 1 compares the two output gap estimates. The substantial factor payments associated with the activities of the multinationals create a sizeable difference between GDP and GNP. For the reasons discussed in Chapter I, GNP may provide a better measure of value added accruing to Irish factors of production than GDP which includes large recorded returns of multinationals. As seen in Figure 1, the differences are considerable and are usually more pronounced during the peaks and troughs of each cycle.

Figure 1.
Figure 1.

Ireland: Output Gaps

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A002

Source: Staff Estimates

8. The estimates obtained with the HP filter are based on a value of the parameter λ equal to 100. The table finally illustrates a range of estimates produced by the European Commission (1999). These reveal how the different methods produce considerably different results. In particular, the estimates obtained by the HP filter imply larger output gaps compared with the production function approach and the unobserved component model.

9. The middle panel of Figure 1 compares the output gaps obtained by the HP filter and the production function approach. As a result of the rapid growth in labor supply and investment in recent years (especially in 1998), the production function approach traces better the growth in potential and is not affected, to the same extent as the HP filter, by the most recent observation. The production function approach tells us what we expect to find out: given that many more workers entered the labor market recently, the economy’s potential must have increased considerably. That simply means that the economy can sustain a higher growth rate without the cost of extra inflation. The results from the production function approach are discussed in the next section.

10. One serious shortcoming of the HP filter is the sensitivity of the derived trend to the end-point. This is illustrated in the bottom panel of Figure 1. In view of the fact that output growth is projected to slow down considerably in 1999, using that year as an end-point results in a positive output gap of 3.8 percent of potential for 1998. This is compared with an output gap of 5.7 percent obtained by the HP filter with 1998 as the end-year. This of course illustrates the dilemma that the policymakers have to confront when making policy decisions in real time based on uncertain estimates of the output gap and potential output growth.

11. In these circumstances it is worthwhile to examine closely the sources of the variation in potential output using the production function approach. In contrast to the statistical smoothing methods, the production function approach provides estimates of potential output using a standard structural production-based approach which allows prior knowledge or expectations about the future evolution of the factors of production to be utilized: it also permits assessment of the respective contributions of the different factors of production and other influences to potential growth.

C. Supply-Side Estimates—Production Function Approach

12. The database used in Magnier (1996) and Samiei and Magnier (1998) was extended with recent data from the IMF and OECD. The total capital stock of the economy was estimated over the period 1970–98 from data on the volume of total gross fixed capital formation over the years 1960–98, assuming a capital-output ratio equal to 2 in 1960 and a constant rate of real depreciation equal to 8 percent afterwards.17 In the absence of an obvious, reliable candidate to represent capacity utilization, it was assumed that it is a function of the deviation of output from a “normal” level. A proxy for this variable is the deviation of actual output from its trend as measured earlier by the HP filter. More details about the estimation are available in Appendix I. Essentially potential output is computed as:

yt*=c+(1α)lt*+αkt+βut*+tfp

where kt and lt denote capital and labor respectively, ut* stands for the supposed “normal” degree of capital utilization and 1* is calculated as:

l*=(nlpar*)(1U*)h*

with 1*par and h* being estimates of the trend participation rate and of the average working hours, and U* an estimate of the standard nonaccelerating wage rate of unemployment (NAIRU).

13. In order to estimate the total trend labor input 1*, the participation rate and the number of hours worked have also been smoothed with the HP filter (λ = 100), and an estimate of U* has been obtained following the simple approach presented by Giorno et al (1995). This method essentially assumes that changes in wage inflation are proportional to the gaps between actual unemployment and the NAIRU:

Δ2w=a(UU*)

where w is an index of nominal wages. With the additional assumption that the NAIRU does not significantly change from one year to another, the NAIRU can thus be simply approximated by:

U*=U(ΔU/Δ3w)Δ2w

and the resulting series are then smoothed with the HP filter to eliminate erratic movements.

14. We present two different scenarios based on different assumptions about labor force growth, investment, wage growth, and unemployment. The assumptions behind the estimation of potential GDP and GNP for Ireland are shown in Table 2 and in Table 5. The baseline scenario is based on a more optimistic set of assumptions about labor input, wage inflation, and unemployment. Under the alternative, less optimistic scenario, it assumed that wage growth accelerates somewhat in 2000–01 and unemployment grows slightly after 1999 following the recent decline. At the same time labor is still expected to grow, albeit at a slower rate compared with 1998.

Table 2.

Baseline Scenario Assumptions

(Percent growth)

article image

Numbers in parentheses are the GNP growth rates.

Baseline scenario

15. Table 3 and Table 4 show the contribution to the growth of potential GDP and GNP (Pot) from labor, capital, and total factor productivity, according to the baseline scenario. It is assumed that female participation rates will to rise further while inward migration flows will continue in the next few years, ensuing a rapid growth in labor (see also OECD, 1999). Notice that changes in labor are the result of changes in unemployment (Emp), growth in the labor force and participation rates (Lab) and hours worked.

Table 3.

Contributions to Potential GDP Growth—Baseline Scenario

article image
Table 4.

Contributions to Potential GNP Growth—Baseline Scenario

article image
Table 5.

A Less Optimistic Set of Assumptions

(Percent growth)

article image

Numbers in the parentheses are GNP growth rates.

16. The estimates show that the growth of potential has accelerated from about 5½ percent in 1993 to nearly 8 percent by 1998. This performance can be explained by the rapid growth in labor input, the strong productivity performance, and, to a lesser extent, high investment flows. A striking result is that most of the pickup in trend growth in the 1990s is accounted for by increases in the labor force rather than total factor productivity. The growth of labor input accounted for about 40 percent of the growth in potential in 1998. Since 1993 the contribution of labor seems to have increased considerably from 1½ percent annually to about 3 percent in 1998, with most ofthat increase coming from an increase in the labor force rather than hours worked. To a lesser extent, the contribution of total capital has also increased from ½ percent in 1993 to 1½ percent in 1998, reflecting high investment flows. Total factor productivity growthhas remained strong but stable throughout the 1990s, after the rapid acceleration posted in the 1980s.

17. Figure 2 shows the actual, potential GDP and the output gap obtained by the production function approach. Based on these estimates the GDP was estimated to be about 2.8 percent above potential in 1998. With these assumptions and an additional hypothesis about TFP growth in 1999–2003, it is estimated that this gap will narrow only slightly to about 2.1 percent above potential in 1999.18 As GDP growth slows down to about 5½–6 percent, the gap is estimated to turn negative, owing mainly to strong growth in potential output.

Figure 2.
Figure 2.

Ireland: Actual, Potential GDP and the Output Gap (Production Function Method)

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A002

Source: Staff Estimates

18. When GNP is used to estimate potential, total factor productivity appears to be somewhat lower (3¼ percent), and as a result the growth of potential output is also lower. In this case the estimates suggest that the GNP was 1 percent above potential in 1998 and will be close to potential in 1999. The gap will turn negative after 2000.

A less optimistic scenario

19. Given the uncertainties in predicting inward migration flows and female participation rates, in the second less optimistic scenario the labor force growth and the rate of growth of working age population are assumed to slow down significantly (see Table 5 for the underlying assumptions). As explained above, under this scenario wage inflation is set to accelerate somewhat while unemployment will rise somewhat after 1999. Table 6 and Table 7 present the contribution of labor, capital and total factor productivity to the growth of potential GDP and GNP respectively.

Table 6.

Contributions to Potential GDP Growth—A Less Optimistic Scenario

article image
Table 7.

Contributions to Potential GNP Growth—A Less Optimistic Scenario

article image

20. Despite a considerable slowdown in the growth of labor after 1998, potential GDP growth continues to be strong, owing mainly to strong TFP improvement. Nevertheless, for a given GDP growth, the growth of potential is about 1 percent lower than in the baseline case. Under these circumstances the output gap in 1998 is estimated to be 3 percent and it is expected to widen further to 3¼ percent of potential in 1999 and 3½ percent in 2000.

Alternative TFP estimates

21. Chapter I provides an alternative estimate for TFP growth based on sectoral industry data from the manufacturing sector which exclude large repatriated earnings by multinationals. This suggests a 2 percent growth in TFP over 1991–96, which is significantly lower than the Solow residual-based estimate derived in this chapter. Using these TFP growth numbers, Table 8 shows alternative estimates for potential output growth over 1999–2000, which avoids some of the measurement problems in aggregate output and income data.

Table 8.

Potential Output Growth With Alternative TFP Estimates

article image

TFP growth based on estimates provided in Chapter I.

22. The estimated growth of potential is about 6¼ percent over 1999–2003. A less optimistic set of assumptions (Table 5) results in a growth rate for potential output of about 5½ percent.

Explaining the growth in total factor productivity

23. Why has TFP growth been so strong in the last few years? And furthermore is this growth likely to persist into the future? It is clear that TFP growth in Ireland, even accounting for measurement problems, has been higher than that in trading partners, especially within EMU. We focus on two possible explanations: first, improved education which affects the overall quality of human capital and, second, the shift from low productivity agriculture to high productivity manufacturing production.

Education

24. Two forces have contributed to a marked improvement in the qualification of new entrants in the labor force. The first stems from the recent demographic trends in Ireland: a late baby boom that got underway in the 1970s peaked in 1980, resulting in a hump-shaped distribution of the population with a peak in the 15–19 age group (see OECD, 1999, for example). This development took place much later compared with other European countries. In absolute numbers more educated young workers have therefore entered the labor market recently and will continue to do so in the next few years.

25. Second, the composition of the Irish labor force has experienced a substantial changeover in the recent years, as the proportion of recent cohorts entering the work force with no qualification has markedly declined. In fact, given that the younger workers entering the labor market tend to be more educated (or have better training), the overall level of education of the labor force has risen considerably.

26. Table 9 shows that 66 percent of those in the age group 25–34 had completed secondary education in 1996 compared with 30 percent in the age group 55–64. A similar gap exists for those with university education, as seen in Table 9. These differences are evident for the OECD area as a whole, albeit not as marked as in the case of Ireland, and highlight the transformation that is underway in the labor market.

Table 9.

Educational Attainment

(Percent of population)

article image
Source: OECD (1999).
Migration

27. Migration affects the education distribution of the population directly. Unlike newly born persons, migrants come with accumulated human capital and thus influence directly the educational distribution of the labor market. Migrants tend to save more and thus contribute growth via higher investment in addition to raising the educational profile of the labor force (see also Barro and Sala-I-Martin, 1995, for example and references therein). Finally, the majority of recent immigrants are of Irish origin and have thus integrated easily with the local workers.

Shift to more productive production

28. The gradual shift of human and capital resources toward the highly productive modern sector associated with foreign direct investment has played an important role in enhancing TFP growth in Ireland. Embedded in the pioneering work of Balassa and Samuelson is the idea that the shift from agricultural to manufacturing production leads to an increase in the measured TFP growth. This is basically due to the fact that TFP growth in manufacturing is higher than in the agricultural sector and therefore the aggregate TFP growth will remain high so long as the economy shifts resources away from the agricultural sector.

D. Output—Inflation

29. Although potential output is defined as that level of output which, if attained, does not set off inflationary pressures, information on price inflation is usually omitted from the estimation19. In this section information on price inflation is used to estimate potential output20. Figure 3 shows the estimated output gap for Ireland based on a model that uses information on the change in CPI to estimate potential. Essentially (the growth of) potential output is modeled as a latent stochastic trend and deviations from trend are linked to inflation through a simple Phillips curve. Thus output and inflation form a bivariate unobserved-components model estimated by maximum likelihood (see Kuttner, 1994, for example). The advantages of using this method are that (i) it allows information on inflation to be used in the estimation of potential output, and (ii) it allows the estimation of uncertainty inherent in the estimation of potential output.

Figure 3.
Figure 3.

Ireland: Output Gap Estimated Using Information about Inflation

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A002

Source: Staff Estimates.

30. Inflation in Ireland has remained subdued in recent years, although output growth has been accelerating. It would appear therefore that the potential of the Irish economy has increased. Again, contrary to the output gap derived by the HP filter, this method reveals that the output gap has closed somewhat between 1996 and 1998. Figure 3 makes it clear that the output gap has not widened, as suggested by the HP filter. The results obtained by this method are closer to the production function estimates. Both use relevant information (about factor inputs, CPI), that the HP filter omits, thereby producing a somewhat smaller output gap.

E. Convergence in the Solow-Swan Model: The Catch-up Hypothesis

31. The neoclassical growth model provides a useful benchmark for motivating the discussion for the medium term growth prospects for Ireland (more details can be found in the Appendix II). The model has interesting implications for Ireland since it predicts that economies with lower capital per person tend to grow faster in per capita terms. In other words it suggests that there will be convergence across economies. This hypothesis, that less rich economies tend to grow faster per capita than rich ones—without conditioning on other country specific characteristics—is referred to as absolute convergence. Indeed there is evidence in favor of convergence especially when similar or homogeneous economies are compared (see Barro and Sala-i-Martin, 1995). Given that Ireland has grown faster than other European countries, Figure 4 shows that the Irish GDP per worker has been converging rapidly to the European levels. A corollary of this hypothesis is that growth will slow down in the medium term and the potential of the Irish economy to carry on growing rapidly without significant wage and price inflation could diminish.

Figure 4.
Figure 4.

Ireland: GDP per Worker Relative to Selected Countries 1/

(In Percent)

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A002

Source: Staff Estimates1/ Ireland’s GDP in PPP dollars per worker, relative to country i.

32. The Solow-Swan model implies gradual convergence to a steady state. For given shares of capital and labor it is possible to make some quantitative estimates for this convergence rate which is essentially a function of the distance between output and the steady state (details in Appendix II).

33. Table 10 shows the growth rates based on two different levels of output relative to steady state (y/y*)21. In the upper panel of the table it is assumed that this ratio is 50 percent while in the lower panel this is set equal to 70 percent. The growth of labor, assumed to be equal to 3.1 percent, is the average growth rate over 1997–2003 and is based on the assumptions from the baseline scenario (Table 2); we also use an alternative estimate of 4 percent growth for comparison. The TFP growth is set equal to 3.5 percent, as estimated by the Solow decomposition, and 2 percent consistent with the alternative estimates provided in Chapter I.

Table 10.

Growth Rates Implied by the Solow-Swan Model

article image

34. The Solow-Swan model implies growth rates in the range of 4–5 percent, which are slightly lower from previous estimates. However, these are not directly comparable with growth rates estimated in the previous section and furthermore, these growth rates are averages of diminishing growth rates as predicted by the Solow-Swan model. Nevertheless, according to this model convergence to steady state is quick. For example, when we assume a rate of technological progress of about 2 percent and a growth of labor of 3.1 percent, the convergence coefficient (defined in Appendix II as β*) implies that 8.5 percent of the gap between y and y* vanishes every year and it takes about nine years to close half of the distance between y and y*.22

F. Concluding Remarks

35. The main focus of this paper was to estimate the potential output for Ireland and evaluate the growth prospects for the medium term. The preferred method for estimating potential output is the production function approach that decomposes the potential into changes in the capital stock, labor, and productivity. The results show that potential output increased considerably in recent years in line with an expanding labor force and strong investment growth and supported by strong productivity growth. Labor supply growth has been the most important factor explaining the pickup in trend growth since the early 1990s and may explain the relatively modest wage growth in Ireland for much of the 1990s.

36. An alternative method, that uses information about CPI growth directly, supports the finding that potential output has increased in recent years. The results obtained by this method are closer to the production function estimates. Both use relevant information (about factor inputs, CPI), that the HP filter omits, thereby producing a somewhat smaller output gap.

37. These estimates of potential output growth suggest that the increase in potential output will slow down eventually as labor force growth levels off, the demographic transformation—which affects the educational characteristics of the population and thus total factor productivity growth—is completed, and investment flows moderate. In addition, accounting for some of the measurement problems in the aggregate output and income data, is shown to reduce significantly the estimated potential growth over the medium term.

APPENDIX I The Production Function Approach

38. This follows closely Magnier (1996) and is based on Giorno et al (1985). The methodology for estimating potential output is based on the standard Cobb-Douglas production function:

yt=c+(1α)lt+αkt+but+tfp+e(1)

where lower case letters denote that the variables are in (natural) logs and lt is the labor input, kt the capital stock, tfpt is the trend total factor productivity (log index), and utt is a proxy for the intensity of use of capital and employed labor (log index). Therefore (1−α) is the elasticity of output with respect to labor and b is the elasticity of output with respect to capacity utilization, c is a constant, and e is a random shock.

39. Total labor input is obtained as:

L=(nlpar)(1U)h(2)

where n denotes the working age population, lpar is the participation rate, U the unemployment rate, and h the average working hours.

40. The total factor productivity (TFP) and parameter b (the elasticity of output with respect to capacity utilization) are jointly estimated by an iterative procedure. Assuming that parameter α is well approximated by labor’s share in GDP, the respective contributions of labor and capital to output are computed and subtracted from the observed (log) GDP, the residual being denoted resd0. A first estimate of parameter b is then obtained by regressing this residual variable on a constant, a linear time trend, and the capacity utilization variable:

resd0=r+vt+b0util+e0(3)

41. The component of resd 0, that is not explained by changes in capacity utilization, provides the first approximation of the log index of total factor productivity:

tfp0=resd0b0util(4)

This approximation is then smoothed with the HP filter and substituted for the constant and linear time trend component r + v t in equation (3). This model is then re-estimated, yielding a new estimate of b, and therefore a new estimate of trend total factor productivity tfp, which is again smoothed, and so on. Final estimates for parameter b and for the trend total factor productivity are thus obtained when the procedure finally converges. When implemented over the period 1970–98, with an assumed constant labor share of income (l–α) = 0.6523, the sequential procedure quickly converges to a fixed value for parameter b equal to 0.73.

Potential output is therefore computed as:

yt*=c+(1α)lt*+αkt+βut*+tfp(5)

where ut* stands for the supposed “normal” degree of capital utilization and l* is calculated as:

l*=(nlpar*)(1U*)h*(6)

with l*par and h* being estimates of the trend participation rate and of the average working hours, and U* an estimate of the standard nonaccelerating wage rate of unemployment (NAIRU).

APPENDIX II The Neoclassical Growth Model

42. This section describes the main elements of the Solow-Swan growth model. For more details the reader can refer to Barro and Sala-i-Martin (1995) and references therein. The production function of the Solow-Swan model, Yt=f(K,L,A)=AKtαLt1α, is neoclassical and has the usual properties: (i) constant returns to scale; and (ii) positive but diminishing returns to capital and labor. In this economy income is consumed and saved and s is the saving rate. For a given depreciation rate, δ, the capital stock evolves over time according to K˙t=sYtδKt where K˙ is the growth rate.

43. The steady-state growth rates in this model are exogenously determined and hence the recommendations for long-run growth that one can make are limited. However, the model has very interesting implications for the transitional dynamics. Given that a steady-state capital/labor ratio exists and is unique (and greater than zero), when the capital-labor ratio increases the growth rate of output declines and approaches zero asymptotically. The model explains why growth is higher for low levels of capital stock. Saving a constant share of income s every period results in new investment, which in turn leads to higher output. Subsequently, each addition to capital raises output by smaller amounts, owing to the diminishing returns to capital. An increase in the saving rate will result, however, in a higher steady-state capital stock, and a temporarily higher (per capita) output growth rate, but permanently higher levels of per capita capital and output.

44. With exogenous technological progress the production function, Yt=F(K, L, A(t)), implies an exogenously increasing, at a rate x, average product of capital. In this case the average product of capital is increasing with time and so is the growth rate of the capital stock. This rate of technological progress is equivalent to total factor productivity (TFP) growth.

45. In summary, the neoclassical model predicts that growth will be higher when the capital stock is relatively low and will decline along with new additions to capital until it reaches the steady state. With the Cobb-Douglas production function, the Solow-Swan model provides us with a quantitative measure of the speed of convergence of the capital/labor ratio to its steady state value; it is straightforward to calculate this speed of convergence (β*) as follows24:

β*=(1α)(x+δ+n˙)

This coefficient gives information about the speed with which output per person approaches its steady state, y*.25 26 The growth rate of output is a function of the distance between output and the steady state and is given by: γy(1α)(x+n˙+δ)[log(y/y*)] as derived in Barro and Sala-i-Martin (1995), where x is the TFP growth, n˙ is the labor input growth rate, δ is the depreciation rate, y and y* are the levels of actual output and steady output respectively.27

REFERENCES

  • Adams C. and Coe D., 1990, “A Systems Approach to Estimating the Natural Rate of Unemployment and Potential Output for the United States,” IMF Staff Papers, vol.37, no.2.

    • Search Google Scholar
    • Export Citation
  • Barro, R. and X. Sala-I-Martin., 1995, Economic Growth, McGraw-Hill.

  • Bayoumi, T., 1999, “Where are we Now and Where are we Going? Estimating the Output Gap and the Growth of Potential Output in Japan,” IMF.

    • Search Google Scholar
    • Export Citation
  • Canova, F., 1998, “Detrending and Business Cycle facts,” Journal of Monetary Economics, 41:475512.

  • European Commission, 1999, “Comparison of trend estimation methods,” Brussels, II/346/99/EN-Revision 1.

  • Giorno, C., P. Richardson, D. Roseveare and P van der Noord., 1995, “Estimating Potential Output, Output Gaps and Structural Budget Balances”, OECD Economic Department Working Paper, No. 152.

    • Search Google Scholar
    • Export Citation
  • Kenny, G., 1996, “Economic Growth in Ireland: Sources, Potential and Inflation,” Central Bank of Ireland Bulletin, Autumn.

  • Kuttner, K., 1994, “Estimating Potential Output as a Latent Variable,” Journal of Business and Economic Statistics, vol.12, no.3.

  • Magnier, A., 1996, “Potential Output, the Output Gap and Inflation in Ireland,” IMF Staff Country Report, SM/96/140.

  • OECD, 1999, 1998–1999 Annual Review—Ireland, Paris.

  • Okun, A.M., 1970, The Political Economy of Prosperity, Washington, DC: The Brookings Institution.

  • Samiei, H. and A. Magnier., 1998, “On the Relevance of the Output Gap in Ireland,” IMF Staff Country Report No 98/126.

16

Prepared by Zenon Kontolemis.

17

Trend changes in the resulting series thus appear similar to those of the business capital stock estimated by the OECD (Economic Outlook data base). The conclusions are not sensitive to any of these assumptions.

18

For the years 1999–2003 it was assumed that the TFP growth will continue to be strong. This is equal to the average growth rate of TPF between 1993 and 1998.

19

An exception is the methodology of Adams and Coe (1990).

20

The data used in this estimation is quarterly (from the OECD database) and consequently differs slightly from the annual data used elsewhere in the study.

21

These are the average growth rates over the period.

22

These estimates are based on the implicit assumption that the saving rate (and therefore investment) is constant. Allowing for a higher rate of investment (financed through foreign direct investment flows) would affect the steady state of the economy and therefore the growth rates.

23

National accounts estimates indicate large changes in the labor share in income since the 1960s. Total labor income measured as the remuneration of employees represented 62 percent of net domestic product at factor cost on average over 1990–94, and 70 percent including income from self-employment and other trading income.

25

This convergence rate does not depend on either the saving rate s for the case of a Cobb-Douglas production function.

26

This is a particular property of Cobb-Douglas technology and is due to two opposite and equal effects (see p.37 in Barro and Sala-i-Martin, 1995, for more details).

27

See Barro and Sala-i-Martin (1995), pp.36–37. This is based on a log-linear approximation of the growth rate around the steady state and it can be used in a recursive procedure to estimate the growth rates during convergence. For the given set of parameters, estimate the speed of convergence (β*) and use the expression for γ y to calculate the growth rate of the economy. This is repeated until convergence to steady state.

Ireland: Selected Issues and Statistical Appendix
Author: International Monetary Fund