Republic of Estonia
Selected Issues

This Selected Issues paper on the Republic of Estonia highlights its growth performance relative to other countries in the European Union (EU). Estonia has experienced a period of unprecedented growth since the mid-1990s. Between 1995 and 2005, Estonia’s real GDP per capita rose by an average of 6½ percent a year, exceeding the annual growth rates of all other countries in the EU. This impressive growth performance is partly explained by the recovery from the immediate post-central planning drop in output.

Abstract

This Selected Issues paper on the Republic of Estonia highlights its growth performance relative to other countries in the European Union (EU). Estonia has experienced a period of unprecedented growth since the mid-1990s. Between 1995 and 2005, Estonia’s real GDP per capita rose by an average of 6½ percent a year, exceeding the annual growth rates of all other countries in the EU. This impressive growth performance is partly explained by the recovery from the immediate post-central planning drop in output.

I. Medium-Term Growth and Productivity in Estonia: A Micro Perspective1

A. Introduction

1. Estonia has experienced a period of unprecedented growth since the mid-1990s. Between 1995 and 2005, Estonia’s real GDP per capita rose by an average of 6½ percent a year, exceeding the annual growth rates of all other countries in the EU. This impressive growth performance is partly explained by the recovery from the immediate post-central planning drop in output. But changes in policies and institutions enhancing catch-up and a favorable global environment have also played an important role.

A01ufig01

GDP per Capita at Constant Prices (PPS), 1996-2005

(Annual changes, in percent)

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Sources: EC (AMECO database); Eurostat; OECD GGDC Total Economy Database; and IMF staff calculations.

2. However, there is still a big income gap with the EU-15 (Figure 1). Although that gap has narrowed since 1995, the Estonian GDP per capita (at current prices and purchasing parity standard, PPS) was only 50.9 percent of the EU-15 average in 2005. Most of the gap stems from a low level of labor productivity. While this is largely due to the paucity of capital, the total factor productivity (TFP) gap relative to the EU-15, at 40 percent in 2005, is also substantial, and catch-up will depend crucially on closing that gap.2

Figure 1.
Figure 1.

GDP per Capita and Productivity Measures, 1995-2005 1/

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Sources: EC (AMECO database); Eurostat; OECD; GGDC Total Economy Database; and IMF staff calculations.1/ The shaded area shows the average for EU-25 (excluding Malta) plus/minus one standard deviation. NMS-8 comprises the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, the Slovak Republic, and Slovenia.

3. These developments raise the question of whether Estonia’s strong growth performance can be maintained over the medium term. To address that question, this paper tries to identify the proximate cause of the recent growth in Estonia and draws some implications for the future. We proceed in two steps. First, we use a growth-accounting methodology based on growth theory to establish some stylized facts at the macro level.3 Since TFP turns out to have been the main engine of Estonia’s growth during the last decade, we then examine the determinants of TFP using micro data. The micro-oriented approach has three advantages. First, since data coverage is better at the micro level, for example on capital stock, micro estimates can be used to cross-check macro findings. Second, micro data give a better picture of the sectoral composition of productivity growth. Finally, micro data provide insights into the firms’ dynamics at play, in particular, restructuring and reallocation, which are critical to understand differences in productivity and growth across sectors and time.4

4. This paper makes several contributions to the literature on growth and productivity dynamics in Estonia. First, we analyze productivity growth from a macro and micro perspective. Second, we use a more recent data set, with a better coverage of the business sector. Third, we match the sectoral distribution of firms with the macro data. Fourth, we improve the estimation of TFP by using the semiparametric approach developed in Levinsohn and Petrin (2003). Finally, we analyze the contribution of different sectors to productivity growth.

5. The rest of the paper is organized as follows. Section B introduces a simple growth-accounting framework and examines Estonia’s growth performance relative to other countries in the EU. Section C analyzes the determinants of TFP growth from a micro perspective. Section D presents a scenario analysis of the impact that changes in specific factors of growth can have on medium-term growth, and Section E concludes.

B. What Do The Macro Data Tell Us?

Growth-accounting framework

6. In order to quantify the contribution of different factors to growth and to the evolution of Estonia’s differential with respect to other economies in the EU, GDP per capita is decomposed into several components. Using a simple identity, we can express output per capita as the product of three components: (1) demographics, (2) labor utilization, and (3) labor productivity:

RealpercapitaGDP=YPop=(NPop)Dem(EN)*(LE)Lab*(YL),Prod(1)

where Y is GDP at constant 1995 prices; Pop is population; Npop

article image
is the ratio of working- age population to total population (inverse dependency ratio); EN
article image
is the employment rate; LE
article image
is the average hours worked per employee; and YL
article image
is labor productivity (i.e., output per hour worked).5 Assuming a Cobb-Douglas production function, labor productivity can, in turn, be decomposed into

YtLt=At*(KtLt)α,(2)

where At represents TFP, a measure of the efficiency in combining a given amount of capital and labor to produce output; KtLt

article image
is the stock of capital per unit of labor, or capital- labor intensity; and the parameter a represents the output elasticity with respect to capital and is set to 0.35.6 Given data for output, capital, and labor for all countries and periods in the sample, we compute TFP (At) for each country as a residual. Hence, TFP encompasses implicitly a variety of factors, such as technological progress, human capital, quality of institutions, etc., that are not captured by the explicitly modeled factors of production—capital and labor.7

How does Estonia compare with other EU countries?

7. The growth-accounting exercise indicates that demographics have made small but positive contributions to growth in Estonia since the mid-1990s (Table 1). In contrast to the EU-15, Estonia’s inverse dependency ratio has increased steadily during the last decade, partly reflecting favorable demographic factors. This has resulted in an annual increase in GDP per capita averaging some 0.3 percent, similar to that of other new member states (NMS).

Table 1.

Sources of Growth 1/

(Average annual percentage change

article image
Souces: EC (AMECO database); Eurostat; OECD; GGDC Total Economy Database; and IMF staff calculations.

Indicators for the EU-15, NMS-8 and Baltics are for the consolidated group (rather than simple averages for the member countries). “Demographics is the working-age population to total population ratio; “labor utilization” is hours worked per working-age person; “employment rate” is the ratio of persons employed to working-age population; “labor productivity” is output per hour worked. GDP and capital stock are valued at 1995 prices and converted to a common purchasing parity standard (PPS) unit of account.

A01ufig04

Inverse Dependency Ratio, 1995-2005 1/

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Sources: EC (AMECO), and Eurostat.1/ Working-age population over total population.
A01ufig05

Fertility Rates, 1970-2005

(In percent)

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

A01ufig07

Employment Rate, 1995-2005

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Sources: EC (AMECO).

8. After declining during the 1990s, labor utilization has also provided an additional boost to growth. As in other NMS, the employment rate fell sharply during the 1990s, reflecting transition-related factors—such as the downsizing or privatization of state-owned enterprises and labor market rigidities—as well as the effect of the Russian crisis in the late 1990s. The fall in employment was compounded by a decline in hours worked and, as a result, labor utilization made substantial negative contributions to Estonia’s growth during the mid-1990s. In this respect, Estonia stands in stark contrast to the EU-15 countries, where the labor input made large contributions to growth during the same period. However, the downward trend of employment and hours worked has been reversed since 2000, thanks to a buoyant economy that has led to stronger job creation than in NMS and EU-15 countries.

9. But the main force behind Estonia’s catch-up in GDP per capita has been labor productivity. Estonia’s labor productivity growth was more than four times that in EU-15 countries and about 75 percent higher than in the NMS during the mid-1990s. The most important factor behind this striking performance was TFP growth, although capital deepening also played an important role (Table 1).8 While the contributions to growth from the capital stock and TFP have eased since 2000, they are still high, particularly compared with EU-15 countries. The lower rise in capital intensity in recent years may be related to a slower substitution of capital for labor, as the flexibility of the labor market has improved, and to the shift toward less capital-intensive sectors, like services. Since the services sector may have absorbed relatively low-skilled workers, this may also explain, at least partly, the slowdown in TFP growth as TFP includes the impact of unmeasured labor quality.

10. What can explain the strong productivity growth during the last decade? One potential explanation is that within-industry efficiency gains—from privatization, greater market incentives, and the adoption of new managerial methods and technologies—increased productivity levels. But shifts in the composition of output toward high-productivity sectors could have also played an important role. To test these hypotheses, we turn to the micro data in the next section.

C. What Is The Micro Story?

Data

11. Our micro data come from the Estonian Business Registry and cover the period 1997-2004. The Business Registry database includes firm-level data from all economic sectors, allowing us to analyze how firm dynamics affect aggregate productivity. Other unique features of the data set include the absence of size thresholds, the availability of transactions data (e.g., mergers), and the provision of detailed information on balance sheets, income, and costs.9 The number of business entities in the registry between 1997 and 2004 more than doubled, and coverage also improved over time. However, because of missing information, we only use the data of about 40 percent of the firms in the registry. Table A.1 presents summary statistics of the variables used in the econometric analysis. Value added and intermediate inputs are deflated by the respective deflators of the system of national accounts provided by the Statistical Office of Estonia. Capital is deflated by the gross capital formation price index. As with most empirical work at the level of aggregation of this paper, the results reported below should be interpreted with caution in light of issues related to data coverage, and measurement and conceptual problems.10

Table A.1.

Summary Statistics 1/

article image
Sources: Estonian Business Registry; and IMF staff calculations.

Employment is expressed in number of workers. The remaining variables are expressed in millions of kroons.

A01ufig08

Number of Firms

(In thousands)

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

12. The coverage of the micro data improves after 2000. The value added of enterprises in our sample accounted for only 45 percent of the aggregate value added in 1997 but increased to 60 percent by 2004.11 Similarly, employment coverage improved over time and was above 50 percent of the macro level in 2004 for all sectors except agriculture and public services, where enterprises are not the main employers. The improvement in data quality may be related to the introduction in 2000 of fines penalizing those firms that do not submit income or balance sheet statements. One drawback, however, is that growth rates in 2000 may be biased because of the improved coverage in that year.12

A01ufig09

Output Growth: Macro and Micro Data

(In percent)

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Sources: Estonian Business Registry database; Statistics Estonia; and IMF staff calculations.
A01ufig10

Micro Data: Employment by Sectors

(In percent of macro employment)

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

13. A preliminary analysis suggests that the business environment has been very dynamic in the recent past. Estonian entry and exit rates are fairly high by international standards, although entry rates have declined over time.13 The high firm turnover may be partly related to the restructuring during the transition period, when there was a shift from large-scale production to smaller units. Also, the relative importance of sectors changed during the sample period (Table A.2). In particular, agriculture has contracted while construction has expanded in recent years. Most firms, however, belong to the services sector, which was underdeveloped during central planning and where smaller firms dominate. Firm size is very small (Table A.3)—about 70 percent of firms are micro enterprises (less than 10 employees)—and is getting smaller across all sectors (Table A.4). In fact, according to Statistical Office of Estonia (2005), the recent increase in the number of enterprises is explained mainly by the birth of micro enterprises.

Table A.2.

Estonia: Number of Firms by Industries

(In percent of total number of firms)

article image
Sources: Estonian Business Registry database; and IMF staff calculations.
Table A.3.

Distribution of Observations Across Size Classes, 1997-2004

article image
Sources: Estonian Business Registry database; and IMF staff calculations.
TableA.4.

Firm Size Across Sectors and Time

article image
Sources: Estonian Business Registry database; and IMF staff calculations.
A01ufig11

Estonia: Firm Entry and Exit Rates

(In percent)

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

A01ufig12

Estonia: Firm Distribution by Ownership

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Methodology

14. We use two alternative methodologies to estimate TFP. Although there is an extensive literature on the empirical identification of production functions, there is some disagreement about the appropriate estimation method. To test the sensitivity of our results to the choice of estimates, we use two methods:

  • Method 1: Industry shares (IS). Assuming a Cobb-Douglas production function, TFP can be calculated as value added minus weighted labor and capital input:

TFPit=logYitαklogKitαllogLit,

(3)

where Yit is the real value added of firm i at time t; Kit is the real capital input; Lit is the labor input (total employment); and α k andαlare the industry cost shares of capital and labor (measured at the two-digit industry level), respectively. Assuming a constant-returns-to scale technology, the capital share is just the residual of the labor cost share, α k=1 -αl.

  • Method 2: Levinsohn-Petrin (LP). Assuming a Cobb-Douglas production function, TFP can be obtain by estimating the equation

logYit=αklogKit+αllogLit+ωit+εit,

(4)

where the error has two components: the unobserved TFP, ω and the error term, s, which is uncorrelated with the input choices. Estimators ignoring the correlation between inputs and unobservable ω will yield inconsistent results.14 To solve for this problem, we follow the semiparametric methodology developed in Levinsohn and Petrin (2003) and use intermediate inputs as a proxy for productivity.15 The main advantage of this method relative to method 1 is that one can test the hypothesis of constant returns to scale since it allows for varying returns to scale across sectors (i.e., α k andαl do not have to add up to one).16

15. To assess the contribution of reallocation and restructuring to productivity growth, we follow Foster, Haltiwanger, and Krizan (1998). Productivity at an aggregated level TFPt at time t is the weighted average of the productivity of individual firms and can be calculated as

TFPt=Σisit*tfpit,(5)

where sit is the output share of firm i in period t and tjpit is the TFP measure. After adjusting the shares Sit to match the sectoral distribution of output in the national accounts data, productivity growth can be decomposed as follows:

ΔTFPt,t1=TFPtTFPt1=Σicsit1*Δtfpit+ΣicΔsit*(tfpit1TFPt1)++ΣiCΔSit*Δtfpit+ΣiNsit*(tfpitTFPt1)ΣiXsit1*(tfpit1TFPt1),(6)

where C, N, and Xdenote continuing, entering, and exiting firms, respectively. The first term in this decomposition represents the within effect, that is, the productivity growth within existing firms keeping the market shares fixed. These productivity gains could be the result of introducing new technological or organizational methods, or of changing the optimal mix of inputs. The second term is the between-firm effect, which reflects productivity growth due to changing output shares. This term will be positive when output shares increase for continuing firms with higher-than-average productivity levels in the previous year. The third term represents a cross (i.e. covariance-type) term. This term will be positive when output shares increase for continuing firms with rising productivity. The last two terms represent the contribution of entering and exiting firms. An entering firm will contribute positively to productivity growth if the firm has higher productivity than the aggregate productivity the year before, while an exiting firm will contribute positively only if the firm exhibits productivity lower than the average in the previous year.

Results

16. The aggregated results from the micro approach are consistent with those based on macro data. In particular, labor productivity has been the main driver of growth at the enterprise level, with TFP accounting for a large share of its variation. Although there is a divergence between productivity growth and value-added growth in 2000, this may be due to the bias introduced by the improved firm coverage in that year. Of the two TFP estimates, LP follows developments in labor productivity closer than IS. Since the assumption of constant return to scale is rejected for most of the sectors, IS probably overestimates the contribution of capital to growth, and, therefore, LP estimates give a better picture.17

A01ufig17

Comparing Productivty Measures

(Annual changes, in percent)

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Sources: Estonian Business Registry database; and IMF staff calculations.

17. The sectoral decomposition indicates that most of the TFP gains are explained by the trade and business services sectors (Figure 2). Average TFP growth between 1997 and 2004 ranged from 0.4 percent (IS measure) to 2.6 percent (LP measure)—compared with a macro estimate of 3.1 percent. Despite this wide margin, the general message from the two estimates is remarkably consistent: both methods indicate that, although agriculture and public services made negative contributions to TFP growth during the period, these were more than offset by large contributions from trade and business services. The industrial sector also made positive contributions to growth but of a smaller scale. The ranking of sectors, however, differs between the two estimates: business services is at the top, according to IS, and trade, according to LP estimates. The performance of the business services is driven by the remarkable TFP growth in the financial intermediation sector, which experienced important changes during the period with the entry of foreign institutions in the Estonian market.18 Within manufacturing, the low-tech and, surprisingly, the high-tech firms exerted a drag on productivity growth.

Figure 2.
Figure 2.

Estonia: Sectoral Decomposition of TFP Growth

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Sources: Estonian Business Registry database; and IMF staff calculations.
A01ufig18

Manufacturing: Average Contributions to TFP Growth, 1998-2004

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Sources: Estonian Business Registry database; and IMF staff calculations.

18. At the same time, firm turnover and reallocation have been the key factors driving TFP dynamics (Figure 3). According to IS estimates the contributions to TFP growth from net entrants have exceeded those from continuing firms for the periods 1998-2004 and 2001-2004. LP estimates shows the same result but only after 2000.19 The positive contribution of firm turnover is due to both entering and exiting firms. In particular, entering firms have on average higher productivity levels than the incumbents had, while exiting firms have lower productivity that continuing firms. But this is far from the whole story. The decomposition of TFP growth also reveals that, although the within and between components of continuing firms have been consistently negative, they have been offset by a large and positive covariance component. That is, aggregate productivity growth has been boosted by the reallocation of output across continuing firms—specifically, firms with rising productivity.

Figure 3.
Figure 3.

Estonia: Decomposition of TFP Growth 1/

(In percent)

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Sources: Estonian Business Registry database; and IMF staff calculations.1/ Contributions to TFP growth are calculated on an annual basis. The figures reported in this graph are average over the corresponding periods.

19. And within each industry, output reallocation has been the main engine of TFP growth (Figure 4). A consistent feature across most of the industries is the importance of the covariance and the negative contribution of the between component. This means that there has been a reallocation of output toward the firms with rising productivity and away from those firms with higher productivity levels. Also, the within component indicates that firms with larger output shares have experienced negative productivity growth. Finally, firm turnover has been important only for the construction and the financial services sectors.20

Figure 4.
Figure 4.

Estonia: Sectoral Decomposition of TFP Growth, 2001-04

(In percent)

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Sources: Estonian Business Registry database; and IMF staff calculations.

20. These findings are subject to several caveats. First, by deflating output and inputs by an industry-level price index we implicitly assume perfect competition. However, under imperfect competition, output prices will differ among firms and, therefore, the firm-level price deviations from the industry-level price will end up in the error term, causing an omitted price variable bias in our estimations. This bias could be potentially important in the financial sector where some firms might have price-setting power. Second, our estimates may suffer from selection bias generated by the relationship between the unobserved productivity variable and the shutdown decision. Third, the covariance effect may be overestimated because measurement errors in output will yield a positive covariance between productivity changes and changes in shares and a spuriously low within-firm effect.

D. Growth Scenarios

21. What do these findings suggest about medium-term growth prospects? We consider three alternative scenarios. The first one assumes three different TFP growth patterns based on historical data. The second one considers the implication of a 50 percent decline in firm entry and exit. Finally, the third scenario analyzes the impact of no further reallocation of output across sectors. All of these scenarios are purely illustrative in nature and should not be confused with projections of medium-term growth.

22. The three scenarios share some common assumptions (Table 2). Population and the inverse dependency ratio are projected to decline over the next 10 years as a result of low fertility rates (see United Nations (2005)). The employment rate is assumed to increase 0.9 percent on average, reflecting the average employment growth rate since 1998. This assumption implies a sharp fall in unemployment (to 2 percent) unless participation rates rise above 73 percent. Hours worked are expected to decline based on income effects and convergence towards lower hours in Europe. However, a further compositional shift of employment toward services or part-time employment could cause a larger decline in hours worked than assumed here. Finally, the capital stock is assumed to increased by 7½ percent per annum over the next 10 years. This is the estimated annual increase in the net capital stock over the period 1998-2005. However, there is scope for faster capital accumulation, given the large capital-labor ratio gap with the EU-15.

Table 2.

Estonia: Growth Scenarios—Assumptions

article image
Sources: United Nations; and IMF staff calculations.

23. The scenario analysis shows that maintaining the strong growth performance of recent years is not a sure thing (Table 3). The first scenario considers three alternative historical patterns of TFP growth: average, best, and worst performance during the period 2001-04. The resulting GDP growth varies within a wide interval—from ½ percent to 13.7 percent. But even if Estonia were to maintain the average TFP growth over the recent past, real GDP growth would slow to about 6 percent. Under the second scenario, TFP growth is dampened by the reduction of firm entry and exit rates, resulting in GDP growth rates below 6 percent. Finally, the third scenario assumes that the output shares of all industries remains fixed at the level of 2004 and TFP growth for each industry equals the average over 2001-04. By keeping output shares fixed, this scenario rules out the reallocation of resources toward the sectors with faster productivity growth, reducing GDP growth to 4-6 percent. These last two scenarios suggest that, a reduction of firm turnover or reallocation could have a significant impact on GDP growth.

Table 3.

Estonia: Total Factor Productivity and GDP Growth: Scenarios

(In percent)

article image
Source: IMF staff calculations.

Based on Industry shares and Levinsohn-Petrin estimates for the period 2001-04.

This scenario assumes a 50 percent decline in entry and exit rates.

This scenario assumes sectoral productivity equal to the average productivity over the period 2001-04 and industries’ output shares fixed at the 2004 level.

E. Conclusions

24. Increases in labor productivity have underpinned the positive performance of the Estonian economy during the last decade. Most of the convergence with the EU-15 achieved during this period stems from closing the gap in labor productivity and, specifically, TFP, which has been boosted by remarkable productivity gains in the business services and trade sectors. Also, productivity dynamics have been dominated by entry and exit. This suggests new firms have introduced technologies and innovation boosting productivity and displaced inefficient firms. At the same time, the reallocation of output toward firms with faster productivity growth has been an important determinant of aggregate and within-sector productivity.

25. But further productivity increases are needed to close the still large gap with respect to advanced economies in the EU. The large TFP gap with respect to the EU-15 underscores Estonia’s substantial growth potential.21 However, strong performance is not guaranteed. Productivity gains in recent years have been driven by firm turnover and reallocation of resources across sectors. But since entry and exit rates have fallen over time and there is a limit to the continuous reallocation of resources, productivity growth might decline in the near future, dampening GDP growth. Therefore, going forward, the main challenge will be to ensure continuous creative destruction and reinvention to move toward (and eventually shift) the technology frontier. Although the assessment of measures to improve productivity lies beyond the scope of this paper, it is worth emphasizing that policies promoting human capital development, R&D investment, the provision of public goods with positive production externalities, and efficient and flexible markets will be critical to foster innovation and rapid productivity growth.

Appendix I. Data Sources and Definitions

Macro data

All macro data come from the Annual Macroeconomic Database (AMECO) of the European Commission unless otherwise noted.

Real GDP: Real GDP at 1995 prices. For comparison purposes, real GDP is converted into a notional currency (PPS) using PPS exchange rates.

Working population: Population aged between 15 and 64 years.

Capital stock: Sum of produced fixed assets that provide ongoing services by being used in the production process for more than one year. For all new member states, capital stock was calculated using the perpetual inventory model with the following inputs: the initial capital-output ratios came from Schadler and others (2006); gross fixed investment was taken from AMECO; and depreciation was assumed to be 5 percent.

Hours worked: Number of hours worked per year and per person employed. Sources: OECD and the Total Economy Database of the Groningen Growth and Development Centre and the Conference Board.

Micro data

The micro data comes from the Estonian Business Registry database. The database covers the period 1995-2004 but, due to missing information, we use data only from 1997. In order to create our sample we follow three steps:

1. Construct a longitudinal panel using registration codes. Several corrections are made to take into account the change in registration codes: (i) firms that change registration codes because of the transfer from the Enterprise Registry to the Business Registry are considered the same firm; (ii) in case of acquisitions, the acquiring and acquired firms are considered a unique firm for the whole sample period; the employment of the acquired firm is added to the employment of the acquiring firm; and (iii) for all other transactions (mergers, breakup, and divesture), we treat firms involved before and after the transaction as different.

2. Exclude unrealistic observations for the variables used to estimate TFP. In particular, exclude individual observations where value added, employment, capital, and intermediate inputs are zero or negative.

3. Exclude those firms for which there is no clear information about the industry they belong to (e.g., because of mergers).

Variable definitions

All variables are in real terms. Value added and intermediate inputs are deflated by the respective deflators of the system of national accounts provided by the Statistical Office of Estonia. The deflators are available for 16 sectors (corresponding to the one-digit ISIC Rev. 3.1). Capital is deflated with the gross capital formation price index.

Value added: Output minus intermediate inputs.

Output: Net sales plus the change in the inventories of final goods.

Employment: Number of employees.

Capital stock: Tangible and intangible fixed assets minus goodwill.

Intermediate inputs: Cost of goods, raw materials, and services purchased for core activities.

Staff costs: Wages and salaries, social security costs, and pension expenses.

Industry classification: Estonian EMTAK code (Classification of Economic Activities of Estonia). Available at: http://www.eer.ee/emtak sisu eng.phtml.

The tables below present some basic statistics about the data.

Appendix II. Levinsohn and Petrin: Methodology and Results

Assuming a Cobb-Douglas production function, total factor productivity can be obtained by estimating the following equation:

logYt=αklogKt+αllogLt+ωt+ɛt=
=αllogLt+φt(Kt,Mt)+ηt,

where M t is the intermediate input and

φt(Kt,Mt)=αklogKt+ωt(Kt,Mt).

Levinsohn and Petrin (2003) find that by substituting a third-order polynomial approximation in Kt and Mt as Σi=03Σj=03iδijKtiMtj

article image
in place of φt(Kt,Mt),
article image
one can consistently estimate the parameters αl and t using OLS. This is the first stage of the estimation procedure. In the second stage , the elasticity of capital ak is defined as the solution to

minαk*Σt(logYtα^llogLtαk*logKtϖt),

where ϖt

article image
is a nonparametric approximationE[ωt|ωt1].
article image
A bootstrap approach is used to construct standard errors for α^l
article image
and α^k
article image
In this study, 50 replications are performed. Once consistent estimates of the input elasticities have been calculated, the log of productivity can be obtained as

ω^t=logYtα^llogLtα^klogKt.

Table A.5 reports the estimated coefficients of the log of labor and capital in the production function of different sectors, as well as the results of the Wald test of constant returns to scale.

Table A.5.

Estimated Coefficients of the Production Function

(Levinsohn-Petrin)

article image
Sources: Estonian Business Registry; and IMF staff calculations.

Wald test of constant returns to scale based on a 5 percent significance.

References

  • Abramovitz, M., 1956, “Resource and Output Trends in the United States Since 1870,American Economic Review, Vol. 46, (No.2),pp. 5– 23

    • Search Google Scholar
    • Export Citation
  • Baily, M. N., and R. M. Solow, 2001, “International Productivity Comparisons Built from the Firm Level,Journal of Economic Perspectives, Vol.15 (Summer), pp. 151– 172.

    • Search Google Scholar
    • Export Citation
  • Bartelsman, E. J., and M. Doms, 2000, “Understanding Productivity: Lessons from Longitudinal Micro Datasets,Journal of Economic Literature, Vol.38, pp. 569– 94.

    • Search Google Scholar
    • Export Citation
  • Foster, L., J. Haltiwanger, and C. J. Krizan, 1998, “Aggregate Productivity Growth: Lessons from Microeconomic Evidence,NBER Working Paper No. 6803 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Girma, S., and Y. Gong, 2007, “FDI, Linkages, and the Efficiency of State-Owned Enterprises in China,Journal of Development Studies (forthcoming).

    • Search Google Scholar
    • Export Citation
  • Griliches, Z., and H. Regev, 1995, “Productivity and Firm Turnover in Israeli Industry: 1979-1988,Journal of Econometrics, 65 (No. 1), pp. 175– 203.

    • Search Google Scholar
    • Export Citation
  • Griliches, Z., and J. Mareisses, 1998, “Production Functions: The Search For Identification,in Econometrics and Economic Theory in the Twentieth Century: The Ragnar Prisch Centennial Symposium, (Cambridge: Cambridge University Press).

    • Search Google Scholar
    • Export Citation
  • Haltiwanger, J, 1997, “Measuring and Analyzing Aggregate Fluctuations: The Importance of Building from Microeconomic Evidence,Federal Reserve Bank of St. Louis Economic Review, May/June, pp. 55– 78.

    • Search Google Scholar
    • Export Citation
  • Hulten, C. R., 2001, “Total Factor Productivity: A Short Biography,in New Developments in Productivity Analysis, ed. by C.R. Hulten, E.R. Dean and M. J. Harper, (Chicago: University of Chicago Press).

    • Search Google Scholar
    • Export Citation
  • Levinsohn, J., and A. Petrin, 2003, “Estimating Production Functions Using Inputs to Control for Unobservables,Review of Economic Studies, Vol.70 (April),pp. 317– 41.

    • Search Google Scholar
    • Export Citation
  • Masso, J., R. Eamets, and K. Philips, 2004, “Firm Demographics and Productivity Dynamics in Estonia,University of Tartu Faculty of Economics and Business Administration Working Paper No. 25 (Tartu, Estonia: Faculty of Economics and Business Administration)

    • Search Google Scholar
    • Export Citation
  • Musso, A., and T. Westermann, 2005, “Assessing Potential Output Growth in the Euro Area: A Growth Accounting Perspective,ECB Working Paper No. 22 (Frankfurt: European Central Bank).

    • Search Google Scholar
    • Export Citation
  • Olley, G. S., and A. Pakes, 1996, “The Dynamics of Productivity in the Telecommunications Equipment Industry,Econometrica, Vol.64 (No. 6),pp. 1263– 97.

    • Search Google Scholar
    • Export Citation
  • Schadler, S., and others, 2006, “Growth in Central an Eastern European Countries of the European Union,IMF Occasional Paper No. 252, (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Statistical Office of Estonia, 2005, Financial Statistics of Enterprises, (Tallin: Statistical Office of Estonia).

  • United Nations, 2005, World Population Prospects: the 2004 Revision, (New York: United Nations).

  • Vanags, A., and , R. Bems, 2005, “Growth Acceleration in the Baltic States: What Can Growth Accounting Tell Us?Baltic International Center for Economic Policy Studies.

    • Search Google Scholar
    • Export Citation
1

Prepared by Marialuz Moreno-Badia. I thank Larissa Merlukova and Kadri Rohulaid of the Centre of Registers and Infosystems for the data and valuable clarifications on the Registrar’s Office database.

2

Estonia’s capital-labor ratio is only about one-third of that of the EU-15.

3

The recent literature is increasingly using the growth-accounting framework to assess potential growth. For an application, see, for example, Musso and Westermann (2005).

4

Empirical papers highlighting the relevance of the connection between aggregate and micro productivity growth include Baily and Solow (2001), Foster, Haltiwanger, and Krizan (1998), Haltiwanger (1997), Olley and Pakes (1996), and Griliches and Regev (1995).

5

For a detailed description of the data and definitions, see Appendix I

6

This is the value used in Schadler and others (2006) and is adopted here to facilitate international comparisons.

7

As Abramovitz (1956) put it, TFP is a measure of our ignorance. Because it is a residual, it includes unwanted components like measurement errors, omitted variables (such as the quality and utilization of capital and labor), and model misspecification. For a review of the literature on TFP, see Hulten (2001).

8

To the extent that the capital stock is underestimated or the gray labor market is large, TFP is overestimated. However, it is widely accepted in the literature that TFP growth has accounted for a large share of labor productivity and growth in Estonia (see, for example, Schadler and others (2006) and Vanags and Bems (2005)).

9

For a detailed description of the data and definitions used in this paper, see Appendix I. For more details on the data set, see Masso, Eamets, and Philips (2004).

10

For example, our sample does not include data for the two major banks in Estonia. Therefore, results concerning the financial sector could be biased.

11

Comparability between the micro and the macro data is limited, however, owing to methodological inconsistencies. Value added at the macro level is a broader concept since it covers not only the activities of enterprises but also of other economic units. According to the Statistical Office of Estonia, all enterprises registered in Estonia accounted for about 70 percent of aggregate value added in 2005.

12

In fact, discrepancies between the value-added growth rates at the micro and macro level widen in 2000.

13

Entrants are defined as those firms for which we have data in period t but not in period t-1. Exiting firms are those for which we have data in period t-1 but not in period t. As a result, the entry and exit rates reported in this paper may be overestimated because of missing data. In any case, Masso, Eamets, and Philips (2004) also find similarly high entry and exit rates using a definition that controls for some of the problems related to missing data.

14

Profit-maximizing firms should respond to positive productivity shocks by expanding output, which requires additional inputs, and vice versa. Therefore, estimating equation (4) by ordinary least squares (OLS) will yield inconsistent estimates. See Griliches and Mareisses (1998) for an overview of the discussion on this subject.

15

Equation (4) is estimated for 40 different sectors (measured at the two-digit industry level). For a detailed description of the semiparametric approach of Levinsohn and Petrin and estimation results, see Appendix II

16

Girma and Gong (2007) have found that the LP and the translog production function methods are superior to the industry shares approach.

17

Table A.5 reports LP estimates and results of the Wald test of constant returns to scale.

18

It could be argued that foreign institutions may have boosted productivity in the financial sector by bringing in more advanced technologies and organization.

19

LP estimates suggest that, on average, continuing firms made larger contributions to growth during 1998-2004. However, we should interpret this result with caution. First, data quality improves after 2000 and, therefore, estimates for the period prior to that may be biased. Second, the net entry effect depends on the horizon over which productivity growth is measured. Other studies on productivity dynamics in Estonia have focused on longer horizons that the one year considered in this paper (e.g., Masso, Eamets and Philips look at 2, 3 and 5 years horizons). As pointed out by Foster, Haltiwanger, and Krizan (1998), studies that focus on high frequency variations (like this paper) tend to find a smaller contribution of net entry to aggregate productivity growth.

20

The net entry in this decomposition is not comparable to the one defined at the aggregate level since entry and exit refer to the specific industry under consideration. For example, a firm that merges with another one and changes the main sector of operation is considered to have exited the industry where it was operating before the merger, even when it has not actually closed down.

21

In addition, higher employment rates and faster capital accumulation could also help maintain rapid income convergence.

Appendix I. Methodology for Making Age-Related Expenditure Projections

Age-related (or person-related) expenditures can in simple terms be expressed as expenditures, E, equaling a benefit level, b, times the numbers of recipients/users, R:

(1)E=b.R

When making long-term projections it is important to identify the underlying (demographic) trends that change the number of recipients and the benefit level. To better understand these trends the expenditure equation can usefully be extended in the following way:

(2)E=bRPPP1564P1564LL

Where P reflects the potentially eligible population for a particular expenditure item (health care, pensions etc.), P1564 is the working age population and L is employment. The eligible population could be total population, the population older than 65 or some other relevant grouping and, thus, the ratioRP

article image
can be understood as the “coverage ratio” of the given expenditure item. The ratioPP1564
article image
is the dependency ratio andP1564L
article image
is the inverse employment ratio. For projection purposes it is assumed that the benefit level is linked to nominal GDP, Y, per worker (as a proxy for the wage rate):

(3)b=br(YL)

This assumption does not preclude the projection taking into account indexation rules that differs from the above relation as this would be captured in the benefit ratio, br. However, the assumption implies that expenditures will stay constant as a share of GDP if there is no change in relative prices, the population composition is unchanged and economic policies do not change the benefit ratio, the coverage ratio and the employment ratio. In the current projections we assume that the benefit ratio will increase for pension expenditures due to ad-hoc decided benefit hikes, while it is also rising for health care expenditures following the effect of non-demographic drivers. Incorporating the relation for the benefit ratio into (2) and taking log-change yields the following equation for the growth in age-related expenditures as a share of GDP:

(4)Δlog(EY)=Δlog(br)+Δlog(RP)+Δlog(PP1564)Δlog(LP1564)

This general equation allows us to undertake long-term expenditure projections taking into account changes in demographics and economic policies. In case of Estonia, the eligible population is assumed to be persons older than 65 when projecting both pension and health care expenditures. The projected number of recipients for pension takes into account the change in the statutory retirement age, while the projected index of health care users is based on the age distribution (five year cohorts) of health care spending (includes the entire population).

Appendix II. Labor Force Projection Using the Cohort Methodology

Rather than assuming unchanged age-specific participation ratios for the full projection period, the cohort methodology allows for changing participation ratios in line with observed exit and entry rates in each five-year cohorts. Due to socio-cultural factors and individual characteristics younger women’s participation rates tend to be higher than older women’s, leading to an autonomous increase in female labor force participation in the years ahead, when the cohort projection methodology is used. In Estonia this positive cohort effect is projected to raise female labor force participation by about 6½ percentage points, while a negative cohort effect for men is projected to lead to a decline in their labor force participation by 1½ percentage points, implying that by 2025 women will have higher labor force participation than men (see Figure A1 and A2).

Figure A.1.
Figure A.1.

Age-Specific Labor Force Participation Rates

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Source: The International Labor Organization (2007), United Nations (2007), and own calculations.
Figure A.2.
Figure A.2.

Aggregate Labor Force Participation Rates

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

Appendix III. The Global Fiscal Model

Model overview

The IMF’s Global Fiscal Model (GFM) is a four-country dynamic general equilibrium model based on the New Open Economy Macroeconomics (NOEM) tradition, but designed to examine fiscal policy issues.37 It is particularly suitable for studying temporary or permanent changes in taxes or expenditures, whether implemented rapidly or occurring gradually (as is the case of age-related expenditure pressures). The GFM analyzes the impact of fiscal policy on real activity through both aggregate demand and supply channels. Aggregate demand responses result from the absence of debt-neutrality and consumers’ impatience. Aggregate supply responses arise from the distortionary effects of taxation. The model features marginal payroll taxes on workers that exceed the average rate, which allows for the consideration of the effects of tax base broadening. The GFM extends the NOEM framework by introducing non-Ricardian features via three distinct channels to allow for thorough fiscal policy analysis:

  • Households have finite horizons and as a result, even temporary changes in fiscal policy affect consumption patterns since any offsetting action required by the government’s intertemporal budget constraint is (perceived to be) borne by future generations and there is no bequest motive.

  • Liquidity-constrained households (a fraction of all households) that consume all their disposable income every period and thereby immediately respond to fiscal policy initiatives that change their disposable income.

  • Distortionary labor- and capital taxes affecting incentives to consume and invest.

Other main aspects of the model are:

  • Consumption and production are characterized by constant elasticity of substitution functions. Firms and workers have some market power, so that prices and wages are above their perfectly competitive levels.

  • There are traded and non-traded goods that allow for a bias toward domestic goods in private or government consumption.

  • There are two factors of production - capital and labor - which are used to produce traded and non-traded goods. Capital and labor can move freely between sectors, but are not mobile internationally.

  • Investment is driven by Tobin’s Q with adjustment costs. Firms respond sluggishly to differences between the discounted value of future profits and the market value of the capital stock.

  • Wages and prices are fully flexible. As a result, monetary policy is ineffective.

  • There are two kinds of financial assets, government debt (traded internationally) and equity (held domestically). International trade in government debt implies the equalization of nominal interest rates across countries over time. However, real interest rates across countries could differ because of the presence of non-traded goods and home bias in consumption.

GFM provides a good platform for discussing the relative merits of alternative fiscal consolidation measures and has been applied to several countries.38 The non-ricardian structure of the model implies empirically plausible responses of key macroeconomic variables to changes in fiscal policy. The wide ranging menu of taxes allows a detailed analysis of the composition of adjustment while the strong microfoundations allow to consider the fundamental determinants of the effects of fiscal policy, such as the response of consumers and producers to changes in fiscal policy as well as the sensitivity to the structure of the economy. Finally, as GFM is an open economy model, it allows for the study of fiscal interdependence.

Calibration

The parameters of the model are calibrated to reflect the macroeconomic features of Estonia (Table A1). In particular, the ratios of consumption, investment, government spending, wage income, and income from capital relative to GDP are set to their current values. Similarly, key fiscal variables—revenue to GDP ratios from taxation of corporate, labor, and personal income and consumption tax, as well as government debt and current government spending—have been calibrated to Estonia’s fiscal structure. Also, the calibration reflects the trading patterns between Estonia, the other Baltic countries and the Nordic countries, Europe, and the rest of the world.

Table A.1.

Key Macroeconomic Variables in the Initial Steady State

article image
Source: IMF staff estimates

The calibration of behavioral parameters is based on general microeconomic evidence found in the literature (seeTable A2). 39 These include parameters characterizing real rigidities in investment, markups for firms and workers, the elasticity of labor supply to after tax wages, the elasticity of substitution between labor and capital, the elasticity of intertemporal substitution, and the rate of time preference.

  • The sensitivity of labor supply to the real after-tax wage (Frisch elasticity) is equal to -0.10 in the baseline value.

  • The elasticity of substitution between labor and capital in the production function equals -0.75.

  • The baseline value for the elasticity of intertemporal substitution is 0.40. This parameter describes the sensitivity of consumption to changes in the real interest rate.

  • The wedge between the rate of time preference and the yield on government bonds determines consumers’ degree of impatience and has not been subject to extensive microeconomic analysis. We have set the baseline value of the wedge to 15 percent (corresponding to a planning horizon of 7 years).

  • The baseline assumes that 70 percent of consumers are liquidity constraint (i.e. excluded from participating in financial markets). As these consumers have no wealth, these households consume 40 percent of aggregate consumption.

  • The baseline assumes that the markup over marginal cost in the tradables sector equals 17 percent and in the nontradables sector equals 25 percent.

Table A.2.

Behaviora l Assumptions and Key Parameters in the Initial Steady State

article image
Source: GFM simulations.

Appendix IV. Stylized Example Explaining the Sustainability Indicator

Consider a stylized economy with an initial primary deficit of 1 percent of GDP, an initial debt stock of 90 percent of GDP, a growth adjusted interest rate of 2 percent, and aging costs increasing linearly by 5 percent of GDP during 2005-50 and stay constant as a share of GDP beyond 2050. With the assumed parameters and time horizon an increase in the initial debt stock or the NPV of aging costs by 1 percentage point of GDP reduces the indicator by 0.02 percentage point of GDP.

In this example the indicator produces an adjustment of 6 percentage points of GDP to restore sustainability. The balance component contributes with 1 percentage point of GDP, while the debt component adds 1.8 percentage points of GDP to the indicator. The contribution from the aging component is 3.3 percentage points of GDP, which is smaller than the total expenditure increase because of the gradual phasing of the higher expenditures and the discount factor (See Table A.3).

Table A.3.

Stylized Example of How to Calculate the Sustainability Indicator

(Percent of GDP)

article image

It is worth emphasizing that compliance with the sustainability indicator always stabilizes the debt ratio at a permanently sustainable level. The debt dynamics as a result of applying the indicator can in some instances even be negative, depending on the initial debt stock and primary balance as well as the profile of age-related expenditure increases.

Figure A.3.
Figure A.3.

Debt Dynamics and Aging Costs in the Stylized Example

(Percent of GDP)

Citation: IMF Staff Country Reports 2007, 256; 10.5089/9781451812534.002.A001

References

  • Au-Yeung, Wilson, Jason McDonald, and Amanda Sayegh, 2006, “Australian Government Balance Sheet Management”,NBER Working Paper Series 12302 (Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Barro, R. J., 1979, “On the determination of public debt”,Journal of Political Economy (Vol. 87) pp. 940– 71.

  • Baylor, M., 2005, “Ranking Tax Distortions in Dynamic General Equilibrium Models: A Survey,Working Paper 2005-06 (Ottawa: Department of Finance).

    • Search Google Scholar
    • Export Citation
  • Baylor, M., and L. Beausejour, 2004, “Taxation and Economic Efficiency: Results from a Canadian CGE Model,Working Paper 2004-10 (Ottawa: Department of Finance).

    • Search Google Scholar
    • Export Citation
  • Botman, Dennis, Douglas Laxton, Dirk Muir, and Andrei Romanov, 2006, “A New-Open-Economy-Macro Model for Fiscal Policy Evaluation”,IMF Working Paper 06/45 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Broda, Christian, and David E. Weinstein, 2004, “Happy News from the Dismal Science: Reassessing Japanese Fiscal Policy and Sustainability,NBER Working Paper No. 10988 (Cambridge: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation
  • Burniaux, Jean-Marc, Romain Duvel, and Florence Jaumotte, 2006, “Coping with Ageing: A Dynamic Approach to Quantify the Impact of Alternative Policy Options on Future Labour Supply in OECD Countries”,OECD Working Paper No. 371 (Paris: Organisation for Economic Co-operation and Development).

    • Search Google Scholar
    • Export Citation
  • Department of Finance Canada, 2004, “Tax Expenditures and Evaluations(Ottawa). Available on the Internet at http://www.fin.gc.ca/taxexp/2004/TaxExp04 e.pdf

    • Search Google Scholar
    • Export Citation
  • European Commission (EC), 2004, “Public Finances in EMU, European Economy No. 3/2004”, (Brussels: European Commission).

  • European Commission (EC) 2006, “The Impact of Ageing on Public Expenditure: Projections for the EU25 Member States on Pensions, Health Care, Long-Term Care, Education and Unemployment transfers (2004-2050)”, (Brussels: European Commission).

    • Search Google Scholar
    • Export Citation
  • Jensen, Svend Erik Hougaard, and Soren Bo Nielsen, 1995, Population Ageing, Public Debt and Sustainable Fiscal Policy”, Fiscal Studies (Vol. 16), pp. 1– 20.

    • Search Google Scholar
    • Export Citation
  • Hauner, David, Daniel Leigh, and Michael Skaarup, 2007, “Ensuring Fiscal Sustainability in G-7 Countries”, IMF Working Paper 07/ (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Heller, Peter S., 2003, Who Will Pay?—Coping with Aging Societies, Climate Change, and Other Long-Term Fiscal Challenges (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Heller, Peter S., and David Hauner, 2006, “Fiscal Policy in the Face of Long-Term Expenditure Uncertainties,International Tax and Public Finance (Vol. 13), pp. 325– 50.

    • Search Google Scholar
    • Export Citation
  • International Labor Organization, 2007, “Yearly Data on Total and Economically Active Population, by Age Group”. Available on the Internet at http://laborsta.ilo.org/.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 2005, “Sweden: 2005 Article IV Consultation—Staff Report” (Washington: International Monetary Fund).

  • International Monetary Fund, 2006, “Denmark: 2006 Article IV Consultation—Staff Report” (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • HM Treasury, 2002, “Long-Term Public Finance Report: An Analysis of Fiscal Sustainability”, (London).

  • Laxton,Douglas, and P. Pesenti, 2003, “Monetary Rules for Small, Open, Emerging Economies,Journal of Monetary Economics (Vol. 50), pp. 1109– 52.

    • Search Google Scholar
    • Export Citation
  • Scherer,Peter, 2002, “Age of Withdrawal from the Labour Force in OECD Countries”, Labour Market and Social Policy Occasional Paper No. 49 (Paris: Organisation for Economic Co-operation and Development).

    • Search Google Scholar
    • Export Citation
  • Statistics Estonia, 2007, Data Available on the Internet athttp://www.stat.ee/.

  • United Nations, 2007, “World Population Prospects: The 2006 Revision Population Database”. Available on the Internet at http://esa.un.org/unpp/index.asp?panel=2..

    • Search Google Scholar
    • Export Citation
  • Whitehead, John, 2006, “Facing Fiscal Futures”, Paper presented at the New Zealand Association of Economists Conference, Wellington. Available on the Internet at http://www.treasury.govt.nz/speeches/nzae/2006/nzae-paper-jun06v2.pdf..

    • Search Google Scholar
    • Export Citation
22

Prepared by Michael Skaarup. Many helpful comments and suggestions from Franek Rozwadowski and Mark DeBroeck are greatly appreciated.

23

This result is supported by Barro(1979) and Jensen and Nielsen (1995) who show that tax smoothing can be motivated in terms of efficiency gains.

24

For discussions of (upside) risks to long-term expenditure projections, see Broda and Weinstein (2004), Heller (2003), Heller and Hauner (2006), and Hauner, Leigh and Skaarup (2007).

25

This approach is also used by both international organizations such as OECD and EU and national institutions, see Scherer (2002), Burniaux et al. (2003), European Commission (2006), and Australian Productivity Commission (2005).

26

This should be seen as an average assumption for the entire projection period. In some years it will be higher, while in others lower—or even negative as in the past few years.

27

The projection incorporates the effect of planned income tax cuts prior to the new government’s State Budget Strategy for 2008-11. Incorporating the additional income tax cuts and partly offsetting indirect tax hikes, which are part of the new strategy, would add to the projected decline in the fiscal balance since the Estonian ministry of finance estimates those to have a negative budget impact of about 1 percentage point of GDP.

28

Since this paper aims to assess the fiscal impact of pensions, it models only the first pillar pension system.

29

The figure compares benefit ratios based on public social security systems only. Estonia, as well as many other European countries, also has second and third pillar pension systems. Including those may change the ranking somewhat, although Estonia likely still would be in the lower end.

30

The decline in the benefit ratio (of the first pension pillar) would, however, be partly mitigated by growing pension benefits from the gradually maturing second pension pillar. Thus, the relative income decline for future pensioners would be smaller than implied by the projected benefit ratio decline under the indexation rule.

31

European Commission (2004). A similar approach is used in HM Treasury (2006).

32

Non-age related current expenditures on goods and services, and transfers and subsidies amounted to roughly 16 percent of GDP in 2005. Real expenditures are here defined as nominal expenditures deflated by the GDP-deflator. Since at least half of non-age related expenditures (compensation of employees) can be expected to grow in line with private sector wages the suggested growth path may entail real reductions in other expenditure components.

33

The baseline pension expenditure projection assumes that pension benefits follow wages from 2030. Thus, a change in the pension indexation rule to track wages has no effect in the late adjustment scenario.

34

Without a change in the pension indexation rule, the needed adjustment would be 4½ percentage points of GDP in the early adjustment scenario. This can be compared to the 5 percentage point adjustment in the late scenario.

35

While the simulation provides an indication of the potential asset position it should not be seen as a forecast of the exact outcome. Even small changes in interest rates or the fiscal balance can change the asset path significantly. Nevertheless, it provides an useful indication of the direction of government assets.

36

The effect of an increase in the growth-adjusted interest rate from 1 percent to 1½ percent has the same magnitude, but with opposite sign.

37

See Botman, Laxton, Muir and Romanov (2006) for a detailed description of the GFM.

38

The model has been applied by IMF staff for background work on recent Article IV consultations with Canada, Germany, Japan, the United Kingdom, and the United States.

39

Other structural parameters have been calibrated using evidence from Laxton and Pesenti (2003) and Batini and others (2005).

Republic of Estonia: Selected Issues
Author: International Monetary Fund