Ireland: Selected Issues and Statistical Appendix
Author: International Monetary Fund
Publication Date:
01 Oct 1999
eISBN:
9781451818772
Language:
English
Keywords:
ISCR; CR; liability; productivity growth; goods price; inflation differential; labor-and productivity; traded goods sector; productivity gain; Wages; Inflation; Manufacturing
Download PDF (1.2 MB)

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.

I. Ireland and the Euro: Productivity Growth, Inflation, and the Real Exchange Rate1

A. Introduction

1. Ireland’s growth performance for most of the decade has been exceptional. Annual GDP growth during 1991–97 has averaged over 6 percent, with the manufacturing sector growing in excess of 10 percent per year. Output growth in Ireland continues to exceed the European average by a wide margin.

2. If output continues to grow at anywhere near recent rates, some real appreciation of the exchange rate might be expected as higher productivity in the traded goods sector increases the demand for labor, pushing up wages and non-traded goods prices (the Balassa-Samuelson effect). In a currency union, this real appreciation would be reflected in higher inflation than elsewhere. Therefore, some wage and price inflation over and above the rest of the euro area might be justified as a market response to Ireland’s rapid productivity growth.

3. If so, it would be useful to quantify the possible magnitude of this effect. With this in mind, this paper examines the recent productivity growth in Irish traded and non-traded goods, and provides some rough estimates of the sort of wage and inflation differentials which would be predicted by a Balassa-Samuelson framework under certain growth assumptions for the future. Of course factors other than the Balassa-Samuelson effect will influence wages and the prices, particularly in the short run.2 For this reason, the intent of the paper is not to predict wages and prices, but rather to provide 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.

4. In assessing the recent growth performance, the paper makes the important argument that measured productivity growth in Irish manufacturing in recent years is likely to overstate the actual productivity gains embodied in Irish factors of production. Most of the measured productivity growth is accounted for by sharply rising returns in a handful of sectors dominated by large foreign-owned export-oriented firms. Since these returns are mainly attributed to the intangible assets of multinationals rather than factors of production located in Ireland, they need not lead directly to wage pressures. Once these returns are taken into account, the scope for wage and inflation differentials appears to be fairly modest.

Inflation differentials in a currency union

5. There is some precedent for inflation differentials to persist in a currency union. In U.S. states, where casual observation suggests that both labor and capital is more mobile than within Europe, regional price differences can be fairly large and surprisingly persistent. Using a panel of price indexes in 15 U.S. cities from 1918 to 1995, Cecchetti, et al (1998) find that prices converge rather slowly—the half-life of convergence is estimated at approximately 9 years. Moreover, there is no evidence that the rate of convergence has increased recently. Consistent with this, annual inflation rates in the U.S. cities measured over 10 year intervals were found to differ by as much as 1½ percent.

6. Possible reasons for regional inflation differentials in the United States, in addition to the Balassa-Samuelson effect on non-traded goods prices, include segmented markets, sticky nominal price adjustment, and transportation costs between regions. Disentangling these effects is difficult. In their paper, Cecchetti, et al (1998) found that for each of four major cities the relative price of non-tradeables rose sharply over time, and that real exchange rates between cities fluctuated substantially, even though price divergence between cities for non-tradeable goods did not appear to be any more persistent than for tradeables. In contrast, Parsley and Wei (1995) found that for 51 specific products in 48 cities between 1975 and 1992 differences in non-traded prices were more persistent than for traded goods, with the half-life of price convergence of services nearly four years compared to one year for tradeables. On the whole, there is evidence that meaningful inflation differentials can arise even with a common currency and relatively high factor mobility, and that the Balassa-Samuelson effect may play a role in creating these differentials.

B. Productivity Growth, Inflation, and the Real Exchange Rate in a Balassa-Samuelson Framework

7. The Balassa-Samuelson framework has become the benchmark model for long-run real exchange rate determination.3 In this model, productivity growth in the traded goods sector raises the demand for labor, pushing up wages in all sectors. Slower productivity growth for non-traded goods requires an increase in their relative price to maintain equilibrium in the labor market. Real appreciation, then, will occur if productivity growth in the traded goods sector is faster than for non-traded goods.

8. How much the real exchange rate appreciates depends on how capital-labor ratios respond to productivity gains (Appendix I). If, as is useful for illustration, perfect capital mobility is assumed the real appreciation is given by:

p^N=θNθTa^T−a^N(1)

where p^N is the percent change in non-traded goods prices relative to traded goods (real appreciation), a^T and a^N are productivity growth rates for traded and non-traded goods, and θT and θN are the labor intensities of traded and non-traded goods production.4 Consistent with this, wage growth in the economy will also reflect productivity growth:

w^=a^TθT=p^N+a^NθN(2)

where ŵ is the wage growth measured in traded goods.

9. It is a small step from real appreciation to deriving inflation differentials. Absent a nominal exchange rate instrument, as in a currency union, any real appreciation will take place through higher inflation. Therefore, the inflation differential will depend on a country’s productivity growth in traded goods relative to non-traded goods compared with the same relative growth in the rest of the currency area. Provided the euro area is not so large as to significantly affect world traded goods prices, and again, assuming perfect capital mobility, the inflation differential will be given by:

π−π*=γ(p^N−p^N*)=γ(θNθTa^T−a^N−θN*θT*a^T*+a^N*)(3)

where γ is the share of non-traded goods in the CPI, and the symbol ∗ denotes the rest of the euro area.

10. It is important to highlight that in the Balassa-Samuelson framework, productivity growth brings about a real appreciation through its impact on the marginal, not the average, product of labor in the traded goods sector. Productivity growth which raises the average product but leaves the marginal product unchanged will not increase labor demand, and will not push up wages and non-traded goods prices. As will be made clear in the rest of the paper, this is particularly important for Ireland.

11. To illustrate this, consider a stylized example of measured productivity growth where output in the traded sector is determined by the following production function:

Y=aL

In this case measured productivity, equal to output per worker, will be the same as the marginal product of labor. Any increase in output per worker reflects an increase in marginal product and will, by raising the demand for labor, push up wages and non-traded goods prices.

12. Now suppose that a foreign-owned firm begins producing in Ireland with the production function:

Y=bL

where b>a. Total output will now be determined by:

Y=aL1+bL2

Measured productivity will be a weighted average of productivity in the two sectors:

YL=αa+(1−α)b

where α is the share of labor in Irish-owned firms. It is clear that the entry of the foreign firm has increased measured productivity, but the marginal productivity of the Irish worker remains unchanged at a (Figure 1). In this case the higher measured productivity in the traded goods sector following foreign entry will have no impact on wages and non-traded goods prices, and there will be no Balassa-Samuelson effect. Although this is clearly an extreme example, it nonetheless illustrates that in certain circumstances measured productivity growth can overstate the potential for real exchange rate appreciation.

Figure 1.
Figure 1.

Labor Market Equilibrium

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A001

C. Traded and Non-traded Productivity in Ireland

Understanding the recent performance of Irish manufacturing

13. A quick look at the performance of the Irish manufacturing sector shows why this might be a concern. The manufacturing sector has been the engine of Irish GDP growth throughout the 1990’s. According to data from the Census of Industrial Production, net output per worker in manufacturing grew by an average annual rate of 9.3 percent in real terms during 1991–96 (Figure 2).5 However, the growth in manufacturing has been overwhelmingly concentrated in just a few sectors. Over 80 percent of the growth in output per worker during 1991–96 is accounted for by five sectors which combined employed only 8 percent of the manufacturing labor force in 1991 (Table 1).6 For all other manufacturing sectors, output per worker grew by an average of 4 percent.

Figure 2.
Figure 2.

Ireland: Real Output per Worker in Manufacturing

(annual percentage change)

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A001

Table 1.

Productivity and Wage Growth in Manufacturing

article image
Sources: Census of Industrial Production, Economic and Social Research Institute, and staff estimates

Deflated by the value-added deflator for manufacturing.

Growth in output per worker in all sectors is greater than for the key and non-key sectors due to shifts in output from low-growth to high-growth sectors.

14. Not only do the key sectors account for most of the growth, but all show very high levels of output per worker which set them off from the rest of the manufacturing industry. For example, output per worker for the subsector of food production which includes cola concentrates was almost IRE £800,000 in 1996 (Figure 3).

Figure 3.
Figure 3.

Ireland: Productivity in Manufacturing by Sector

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A001

15. Importantly, only a small fraction of this output goes to pay wages; wages as a share of net output for the key sectors averaged only 8 percent, compared to 35 percent for all other manufacturing sectors. Indeed, even with very strong growth in output per worker, real wages in the key sectors grew by only 1 percent per year during 1991–96, compared with about 3 percent in all other manufacturing sectors.

16. The unusually high returns in the key sectors are not surprising if understood in the context of the global operations of large multinationals which dominate these sectors. Typically, the multinational would have invested a large amount of resources globally in research, product development, and advertising, and would choose the location of its producing subsidiaries based on a number of considerations including, importantly, the tax regime in the host country. If the parent company’s global development expenses are large, high output per worker in its subsidiaries would be required to generate a return to investment on its intangible assets.

17. Something like this explains the output performance in the key sectors in Ireland. The degree of foreign ownership in these sectors is very high: in 1996, 95 percent of the value-added produced in the key sectors was produced by foreign-owned enterprises, compared with 54 percent for all other manufacturing sectors. A large majority of these enterprises’ output is exported. The cola industry, for example, is dominated in Ireland by Coca Cola and Pepsi. The global R&D and advertising expenditures of these companies would be large, and the profitability of the subsidiary operations in Ireland would need to be high to generate a return on investments at headquarters. Understood in this context, it is no surprise that the measured output per worker in the cola concentrates industry is unusually high, and that only a small portion of this output is passed on in the form of wage payments.

18. The important implication is that the very high measured output per worker in Ireland’s key sectors is likely to overstate the marginal productivity embodied in capital and labor physically located in Ireland. Rather than reflecting the inherent productivity of Irish labor, the high output per worker in the key sectors at least partly reflects the intangible assets of multinationals. While there is some evidence that over time foreign investment has permanently raised the productivity of Irish workers as know how is accumulated and ideas and practices spill over into Irish industry (O’Malley, 1998), the measured productivity of workers in the key sectors would almost certainly return to more typical levels if these foreign companies located elsewhere. If this is the case, then Ireland’s recent productivity growth is unlikely to result in wage pressures in the same magnitude, and will overstate the scope for Balassa-Samuelson effects.

Estimating productivity growth in Irish manufacturing

19. To get a better measure of the sort of productivity growth that would lead to a Balassa-Samuelson effect, it would be necessary to strip out that part of measured output which represents a return to intangible assets of multinationals, with the remaining output attributed to its factors of production-capital and labor-and productivity. Measured in this way, productivity would be more likely to reflect changes in the marginal product of Irish labor. By comparing this adjusted productivity growth for traded and non-traded goods with growth rates in euro area economies, a rough estimate of the possible Balassa-Samuelson effects can be derived.

20. Unfortunately, there is no direct way to isolate returns to the intangible assets of multinationals. In principle, data on repatriated earnings, royalties, and other payments of subsidiaries to parent companies could be used, but these are only available in aggregate based on balance of payments data.7 One way to overcome this data shortcoming is to recognize that the key sectors are exceptional with respect to output per worker, and to subtract that portion of net output after payment of wages and salaries (hereafter referred to as “returns”) which is above a certain chosen threshold. The choice of the threshold would be guided by some observed “norm” for the manufacturing industry. This is in essence the approach taken by Conroy et al (1998), who conclude that for many of the key sectors returns were unusually large both with respect to the estimated capital stock and in comparison with similar sectors in EU countries; these excess returns, they argue, reflect the use of technological, scientific or market knowledge, brands, and other elements contributing to market power. While choosing a threshold based on a norm is somewhat arbitrary, the approach is nevertheless sensible once it is recognized that the key sectors in Ireland are indeed outliers because of their access to intangible assets.

21. By all accounts, measured returns in the key sectors have greatly exceeded average returns for all other sectors; during 1991–96, returns as a share of value-added in the key sectors have averaged over 85 percent, compared with 54 percent for all other sectors (Figure 4).8 Returns in have been particularly high for cola concentrates (95 percent), software (92 percent), and basic organic chemicals (89 percent). The distribution of returns across sectors is somewhat skewed: in 1991, over 75 percent of all sectors had returns which fell below a threshold of 53 percent.

Figure 4.
Figure 4.

Ireland: Total Returns as a Share of Value Added 1/

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A001

Source: Census of Industrial Production; and Economic and Social Research Institute.1/ Total returns are defined as value-added less wages and salaries less investment in physical capital.

22. Since little of the growth in output per worker in the key sectors was paid out as wages, it is no surprise that removing excess returns from total output has a profound affect on estimates of total factor productivity growth in manufacturing (see Appendix II for a full description of the estimation approach). On an unadjusted basis, measured productivity is estimated to have grown at an average annual rate of 6.8 percent during 1991–96 compared with 9.5 percent in the key sectors and 2.8 percent in all other sectors (Figure 5 and Table 2). However, when returns above the threshold of 53 percent of value-added are removed the adjusted productivity growth is estimated at only 3.9 percent.9 Although the adjusted growth rate is sensitive to the threshold chosen, for all reasonable thresholds productivity growth is sharply lower when the unusually high returns in the key sectors are taken into account.

Figure 5.
Figure 5.

Ireland: Total Factor Productivity in Manufacturing

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A001

Table 2.

Estimated Total Factor Productivity in Manufacturing.

(average annual percentage change, 1991–96)

article image
Sources: Census of Industrial Production, ESRI, and staff estimates.

Value-added is adjusted by subtracting excess returns above the specified threshold. See Appendix II.

Traded and non-traded productivity in Ireland and the euro area

23. As the largest traded industry, lower estimated productivity growth in manufacturing should reduce measures of productivity growth for traded goods as a whole, and thus the magnitude of the expected Balassa-Samuelson effect. Supplementing the productivity growth in manufacturing (estimated using the 53 percent threshold) with national accounts data for other sectors of the economy, productivity in the traded goods sector during 1991–96 grew by an estimated 3.5 percent compared with a measured growth rate of 6.1 percent (Table 3 and Figure 6).10 This compares with annual productivity growth of 0.9 percent for non-traded goods. Since the productivity growth rates are determined over a period when output was both cyclically high (1994–96) and low (1991–93), they are more likely to reflect trends abstracting from the business cycle.

Table 3.

Ireland: Total Factor Productivity by Sector

(Annual growth rate in percent)

article image
Source: Economic and Social Research Institute, Census of Industrial Production, and staff estimates.

Includes adjustments for financial services, and taxes on expenditure minus subsidies.

Figure 6.
Figure 6.

Ireland: Total Factor Productivity by Sector, 1980–96

(Index 1980=100)

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A001

Source: Economic and Social Research Institute; Census of Industrial Production; and staff estimates.

24. The inflation differential will depend on how these growth rates compare with those in the euro area. Even adjusting for returns to intangible assets abroad, productivity growth for traded goods in Ireland appears to have been significantly above euro area rates in the 1980s and early 1990s. Although data for all euro area countries are not available, comparisons can be made for a subset of euro area countries (including Germany, France, Italy, Belgium, and Finland) during the period 1980–93. Based on a weighted average of OECD country estimates of total factor productivity by sector, annual growth in the traded goods sector for the whole period averaged only 1.3 percent for the euro area countries considered, or 2.0 percent for the faster growth period 1987–92 (Table 4, and Figure 7).11 By comparison, non-traded growth averaged 0.9 percent for the period 1980–93, and 1.2 percent during 1987–92.

Table 4.

Euro Area: Total Factor Productivity by Sector 1/

(Annual growth rate in percent)

article image
Source: OECD and staff estimates.

Includes data for Germany, France, Italy, Belgium, and Finland.

Figure 7.
Figure 7.

Euro Area Total Factor Productivity

Citation: IMF Staff Country Reports 1999, 108; 10.5089/9781451818772.002.A001

D. Estimating inflation differentials in a Balassa-Samuelson framework

25. What do these growth rates, were they to continue, suggest about Irish wage growth and inflation which could be expected in a common currency area? It is worthwhile first to consider the inflation differential assuming perfect capital mobility, and then later to examine the effects if, as is more plausible in the near term, capital adjusts gradually.

26. Table 5 shows four variables for Ireland predicted by the Balassa-Samuelson framework—the inflation differential, the inflation rate, nominal wage growth, and real appreciation of the exchange rate—under various scenarios based on the framework derived in Appendix I.12 As expected, once excess returns in Irish manufacturing are accounted for the inflation differential predicted by the Balassa-Samuelson framework falls substantially; adjusted growth in Irish tradeables gives rise to an inflation differential of VA percent, compared to a differential of almost 3 percent if actual measured growth is used. Predicted nominal wage growth is much higher; even after adjusting for excess returns, wage growth of some 8 percent would be predicted if capital is assumed to adjust fully and instantaneously. However, this assumption is clearly inappropriate, as it implies an annual growth rate in the capital-labor ratio for traded goods (7½ percent) which is implausibly high, and well in excess of the actual rate of 1¾ percent during 1991–96.

Table 5.

Ireland: Inflation, Wage Growth, and Real Exchange Rate Forecasts

in a Balassa-Samuelson Framework

(Annual percentage change)

article image
Source: ESRI, Census of Industrial Production, Central Statistical Office, and Staff estimates.

Assumes the aggregate capital stock grows at the rate of 2 1/2 percent per year.

Assumes total labor supply grows at the rate of 2 1/2 percent per year.

Assumes total labor supply grows at the rate of 1 1/2 percent per year.

Based on an assumed euro area inflation rate of 1 percent per year.

Defined as the percentage change in the price of non-traded goods relative to traded goods

In the case of perfect capital mobility, the capital labor ratio is endogenously determined.

27. Since immediate and full adjustment of the capital stock is not particularly likely in the near term, it is worth considering the predictions of Balassa-Samuelson under alternative assumptions about capital and labor growth. This is particularly important for wage growth, since rapid capital accumulation would imply a higher marginal product of labor, whereas gradual adjustment of the capital stock would dampen the growth in wages.

28. Besides perfect capital mobility, two scenarios for capital-labor ratio growth in Ireland are considered. In both scenarios, growth in the capital stock is assumed to continue at the same relatively rapid rate as during the period 1993–96, about 2½ percent per year. In the first scenario, labor supply is assumed to grow at the relatively rapid rate of about 2½ per year. In the second case, more moderate labor supply growth of about 1½ percent is assumed.13 In both scenarios, euro area capital-labor ratio growth is assumed to continue at its historical rate.

29. It is clear from Table 5 that different assumptions on capital and labor growth do not have much effect on the inflation differential; in all cases, inflation is predicted at just over 1 percent higher than in the euro area. Assuming a euro area inflation rate of about 1 percent, this suggests that Ireland could sustain price inflation rates of about 2 percent over the medium term. Predicted wage growth, however, is substantially lower than in the case of perfect capital mobility. Under more plausible assumptions on the capital-labor ratio, wage growth of some 5 to 5¼ percent per year could be sustained over the medium term.

30. Finally, it is worth noting that the wage and inflation growth predicted by the Balassa-Samuelson framework is relatively insensitive to different assumptions about euro area growth. Although these results are not shown, if euro area productivity growth were to continue at about 1 percent per year rather than 1½ percent assumed in Table 5, both the predicted inflation differential and predicted wage growth would rise by about 0.1 percent per year for each of the scenarios considered.

APPENDIX I A Balassa-Samuelson Framework for Determining Inflation Differentials

31. This section derives the Balassa-Samuelson relationship between productivity and the real exchange rate, and extends this relationship to determine inflation differentials.

The real exchange rate and wages

32. Start with a small open economy producing traded and non-traded goods according to:

YT=AT(LT)θT(KT)1−θT(8)
YN=AN(LN)θN(KN)1−θN(9)

where output (Y) in each sector is a function of labor (L), capital (K), and productivity (A).14

33. For a given distribution of capital, equilibrium in the labor market will be reached when the wage rate equals the value of the marginal product of labor in each sector:

W=θTAT(KT/LT)1−θT(10)
W=PNθNAN(KN/LN)1−θN(11)

where W is the wage measured in traded goods and PN is the relative price of non-traded goods. Equations 10 and 11 jointly determine the relative price of non-tradeables (the real exchange rate) as a function of the capital labor ratio. Log-differentiating both equations and solving for the percentage change in the relative price of non-tradeables gives:

p^N=a^T−a^N+(1−θT)k^T−(1−θN)k^N(12)

That is, the real exchange rate will appreciate with higher productivity growth in tradeables relative to non-tradeables, and with faster capital-labor accumulation in the tradeable sector (accounting for relative labor intensities). The wage growth consistent with equations 10 and 11 is given by:

w^=a^T+(1−θT)k^T=+p^N+a^N+(1−θN)k^N(13)

34. The relationship between the real exchange rate and productivity growth can be simplified by endogenizing the capital-labor ratios. This is done by assuming that with perfect capital mobility the capital-labor ratios are those which equate the marginal product of capital to a world rate of return (R).

R=(1−θT)AT(KT/LT)−θT(14)
R=PN(1−θN)AN(KN/LN)−θN(15)

From equations 14 and 15, growth in the capital labor ratios will depend on productivity growth:

k^T=a^TθT,k^N=p^N+a^NθN(16)

Substituting equation 16 into equations 12 and 13 gives:

p^N=θNθTa^T−a^N(17)
w^=a^TθT=p^N+a^NθN(18)

The inflation differential

35. The price level in Ireland will reflect both traded and non-traded goods prices:

P=(PN)γPT(19)

where P is the price level in euros, PT is the traded goods price in euros, and y is the share of non-traded goods prices in the CPI. As before, PN is the relative price of non-traded goods. From equation 19 inflation will be:

π=γp^N−p^T(20)

Similarly, inflation in the rest of the euro area will be:

π*=γ*p^N*−p^T*(21)

Assuming equal shares of non-traded goods in consumption, and assuming weak purchasing power parity for traded goods ((p^T=p^T*)), the inflation differential will be a function of the relative real appreciation of the exchange rate:

π−π*=γ(p^N−p^N*)(22)

In the case of perfect capital mobility, and assuming that the euro area is not so large as to significantly affect the world price of tradeables, this differential will reflect productivity differences as follows:

π−π*=γ(θNθTa^T−a^N−θN*θT*a^T*+a^N*)(23)

APPENDIX II Estimating Total Factor Productivity in Irish Manufacturing

The framework

36. This section describes the approach taken in estimating total factor productivity growth for Irish manufacturing industries during 1991–96.

37. A rough production function is estimated using data from 130 manufacturing sub-sectors (including mining) taken from the Census of Industrial Production for the years 1991–96. The production function takes the following form:

Yit=AitLitαKitβ(24)

where Yit is the net output (value-added) of sector i at time t, Lit is total number of workers engaged, Kit is the capital stock, and Ait is total factor productivity. In terms of logs, this becomes:

yit=ait+αlit+βkit(25)

38. The coefficients on capital and labor are estimated using OLS. To better estimate the effect of changes in capital and labor on output across sectors and over time, time- and sector-dependent dummy variables are introduced. The estimated production function is therefore:

yit=Cit+αlit+βkit+δ1TIMEt+δ2SECTORt+ϵit(26)

where the sector dummy variables are defined at the 3-digit level using the NACE Rev 1 industrial classification system. Estimated total factor productivity is defined as:

a^it≡yit−α^lit−β^kit(27)

The data

39. Output for each sector is taken to be the nominal net output from the Survey of Industrial Local Units (part of the Census of Industrial Production) deflated by the value-added deflator for manufacturing (provided by ESRI). Although ideally each sector would have its own value-added deflator, sectoral deflators at this level of disaggregation and corresponding to the NACE Rev 1 industrial classification were not available. When productivity is estimated on an “adjusted” basis, output corresponds to total output less returns above a chosen threshold level. Labor is defined as the total number of workers engaged by industrial units, including employees, proprietors, and unpaid family workers.

40. Since capital stock are not reported in the Census, the series used in the estimation was constructed from Census data on investment. Ideally, investment data from previous censuses would be used to calculate a capital stock variable at an assumed depreciation rate.

However, the system of industrial classification was changed in 1991, and no direct correspondence between pre- and post-1991 classification can be made using data aggregated to the 4-digit level.15 To overcome this, a proxy for the capital stock in 1990 was constructed using data on the total capital stock in that year for all manufacturing sectors, distributed among sectors in proportion to the sum of each sector’s investment during 1991–96. While the distribution of the capital stock in 1990 is surely correlated with subsequent investment, using a proxy instead of the actual series will introduce noise into the estimation. Part of this noise will be corrected by including the sector-specific dummy variables, but the remaining noise could lead to an underestimation of total factor productivity in a sector if unusually large capital outlays were made in that sector prior to 1991. The capital stock for the years 1991 was constructed by adding investment (net additions to capital assets), deflated by an investment deflator for manufacturing provided by ESRI, and assuming a depreciation rate of seven percent.

Estimated total factor productivity

41. Results for estimated equation 26 are shown in Table A1. The fit is good, with the combined coefficients on capital and labor implying slightly higher than constant returns to scale. Total factor productivity estimates for each sector are weighted by the sector’s inputs and summed to calculate total factor productivity for manufacturing as a whole. Based on this, average annual total factor productivity growth is also shown in Table A1.

Table A1.

Ireland: Estimated Total Factor Productivity in Manufacturing, 1991–96

article image
Source: Census of Industrial Production, ESRI, and Staff estimates.

42. When using measured output, total factor productivity grew at an average annual rate of 6.8 percent compared with 9.5 percent for the key sectors and 2.8 percent for all other sectors. Adjusted total factor productivity is estimated by subtracting from value-added the portion of returns as a share of value-added above a chosen threshold, where returns are defined as value-added less wages and salaries less investment in physical capital. Three different threshold levels were considered: (i) 47 percent below which half of all sectors had returns as a share of value-added in 1991; (ii) 53 percent below which three quarters of all sectors had returns; and (iii) 69 percent below which 90 percent of all sectors had returns. When the threshold of 53 percent is applied, the estimated productivity growth falls to 3.9 percent. As is clear in Table A1, this growth rate is sensitive to the threshold chosen.

REFERENCES

  • Alberola-Ila, Enrique, and Timo Tyrväinen, 1998, “Is there Scope for Inflation Differentials in EMU?” Bank of Spain Working Paper No. 9823.

    • Search Google Scholar
    • Export Citation

  • Canzoneri, Matthew B., Robert E. Cumby, and Behzad Diba, 1999, “Relative Labor Productivity and the Real Exchange Rate in the Long Run: Evidence for a Panel of OECD Countries,” Journal of International Economics, 47, pp. 245–266.

    • Search Google Scholar
    • Export Citation

  • Cecchetti, Stephen G., Nelson C. Mark and Robert Sonora, 1998, “Price Level Convergence Among United States Cities: Lessons for the European Central Bank,” CEPR Conference Paper, September.

    • Search Google Scholar
    • Export Citation

  • Conroy, C., P. Honohan, and B. Maitre, 1998, “Invisible Entrepot Activity in Irish Manufacturing,” Irish Banking Review, Summer, pp. 22–38.

    • Search Google Scholar
    • Export Citation

  • De Gregorio, Jose, Alberto Giovannini, and Holger C. Wolf, 1994, “International Evidence on Tradables and Nontradables Inflation,” European Economic Review, 38, pp. 1225–1244.

    • Search Google Scholar
    • Export Citation

  • Froot, Kenneth A., and Kenneth Rogoff, 1995, “Perspectives on PPP and Long-Run Real Exchange Rates,” in G. Grossman and K. Rogoff (eds.), Handbook of International Economics, Vol. III.

    • Search Google Scholar
    • Export Citation

  • O’Malley, Eoin, 1998, “The Revival of Irish Indigenous Industry 1987–1997,” Quarterly Economic Commentary, April, pp. 35–62.

  • Parsley, David C, and Shang-Jin Wei, 1996, “Convergence to the Law of One Price without Trade Barriers or Currency Fluctuations,” Quarterly Journal of Economics, 111, November, pp. 1211–1236.

    • Search Google Scholar
    • Export Citation
1

Prepared by Brian Aitken.

2

Given Ireland’s position in the business cycle relative to the rest of the euro area, cyclical factors might be particularly important in explaining short run real exchange rate movements; if growth in Ireland slows and European growth picks up, some pressure toward real depreciation might at least partially offset any long-run Balassa-Samuelson effect. On the other hand, the Irish pound is widely regarded as having locked into the euro at a somewhat undervalued exchange rate, which would lead to pressure for a one-time real appreciation.

3

See Froot and Rogoff (1995) for a discussion of the Balassa-Samuelson effect, as well as the empirical success of the long run relationship between productivity differentials and the real exchange rate. For a more recent empirical treatment, see Canzoneri et al (1999). With regard to inflation, Alberola-Ila and Tyrvainen (1998) study the relationship between productivity differentials and the real exchange rate for 8 of 11 EMU countries, and infer the potential for inflation differentials in EMU. They argue that for several EMU countries inflation differentials of around one percent could be expected.

4

If capital is perfectly mobile, real appreciation is determined entirely by supply side conditions. If capital is not perfectly mobile, as is likely to be the case in the short run, demand side conditions are often modeled. These are discussed in Froot and Rogoff (1995). Since the focus of this paper is on real exchange rate and inflation developments abstracting from the business cycle, demand side conditions are not highlighted. Nonetheless, in estimating plausible Balassa-Samuelson effects, the paper considers the case in which capital cannot be transferred instantly across sectors.

5

Output figures discussed in this section are taken from the Census of Industrial Local Units, and represent net output, or value-added, which is defined as gross output less materials, industrial services, and fuel and power. Mining, which represents a small fraction of industrial output, is included in order to make the data comparable to OECD data for euro area countries analyzed later in the paper. The nominal output figures are deflated using the value-added deflator for manufacturing provided by the Economic and Social Research Institute (ESRI), since sectoral deflators were not available for the new industrial classification.

6

These sectors are “homogenized food preparations” and “other food products” (1588, 1589) including cola concentrates, “reproduction of recorded media” (223) including software, “other organic basic chemicals” (2414) representing mainly pharmaceutical inputs, “pharmaceutical preparations” (2442), and “office machinery and computers” (30). See Conroy, et al (1998).

7

In the balance of payments, factor payments associated with the activities of multinationals are large, and account for much of the substantial difference between GDP and GNP levels as well as growth rates. See Chapter II, Potential Output Growth in Ireland.

8

Here, “returns” are defined as value-added less wages and salaries less investment in physical capital. The investment component is intended to proxy for outlays necessary to offset depreciation of existing capital. Since this component is generally small, it makes little difference if it is excluded. As an alternative to returns as a share of value-added, returns could be measured as a share of the capital stock. Although these two measures turn out to be strongly correlated, the capital stock is a constructed series (discussed in more detail below) rather than a reported value. As such, it is much less reliable at the sectoral level than value-added. For this reason, the norm is determined with respect to returns as a share of value-added.

9

Adjusted Total factor productivity was measured using three different threshold levels: (i) 47 percent below which half of all sectors had returns as a share of value-added in 1991; (ii) 53 percent below which three quarters of all sectors had returns; and (iii) 69 percent below which 90 percent of all sectors had returns.

10

Total factor productivity growth for each sector (other than for manufacturing) was estimated using the residual of an accounting decomposition of output growth based on changes in inputs, (the Solow residual) where the labor intensity for each sector was assumed to equal the average over the period considered of wages and salaries as a share of total output. National accounts data in constant prices and estimates of capital stock by sector are provided by ESRI. Traded goods are defined as manufacturing, mining, agriculture, forestry, and fishing, representing about one third of total output (Table 3). While part of distribution, transport and communications might also be included in the definition of traded goods (see De Gregorio and Wolf, 1994), data restrictions did not allow this sector to be split into traded and non-traded components in the case of Ireland. Given the relatively small size of this sector, this is unlikely to affect the results meaningfully.

11

Data are from the OECD International Sectoral Database. Euro area-wide estimates of productivity growth are constructed by aggregating for each sector OECD total factor productivity estimates for all countries, weighted by each country’s PPP-based U.S. dollar output for the sector.

12

The calculations for nominal wage growth and inflation are based on a euro area inflation rate of 1 percent. As discussed in Appendix I, the precise inflation differential which would be predicted depends on the extent to which traded goods productivity growth in the euro area affects world prices for tradeables. The calculations in Table 5 take as a baseline the case that the euro area growth does not significantly affect world prices. Relaxing this assumption is not likely to fundamentally alter the results.

13

These two scenarios are consistent with the “baseline” and the “less optimistic” growth scenarios considered in Chapter II-Potential Output Growth in Ireland.

14

The model in this section is based on Froot and Rogoff (1995).

15

A correspondence between the two classification systems exists, but mapping the two systems directly requires much more desegregated data than are reported in the Census.

Same Series

Other IMF Content

Other Publishers

Cover IMF Staff Country Reports
Ireland: Selected Issues and Statistical Appendix
Author: International Monetary Fund
Volume/Issue:
Volume 1999: Issue 108
Publisher:
International Monetary Fund
ISBN:
9781451818772
ISSN:
1934-7685
Pages:
111
DOI:
https://doi.org/10.5089/9781451818772.002

Labor Market Equilibrium

Ireland: Real Output per Worker in Manufacturing

(annual percentage change)

Ireland: Productivity in Manufacturing by Sector

Ireland: Total Returns as a Share of Value Added 1/

Ireland: Total Factor Productivity in Manufacturing

Ireland: Total Factor Productivity by Sector, 1980–96

(Index 1980=100)

Euro Area Total Factor Productivity