Hungary: Selected Issues
Author: International Monetary Fund
Publication Date:
25 Oct 2006
eISBN:
9781451818048
Language:
English
Keywords:
ISCR; CR; Hungarian forint; Hungary; EU country; employment rate; employment growth; country yearbook; employment benchmark; Labor costs; Real effective exchange rates; Tax wedge; Employment rate
Download PDF (231.5 KB)

The two legs that have held up the forint in recent years—a strong “EU accession effect” and positive sentiment toward emerging markets—may no longer be strong enough to offset Hungary’s weak fundamentals. Fiscal consolidation efforts should be supported by stronger budget controls and greater transparency and accountability. This paper is an effort to shed light on Hungary’s employment dynamics, placed in the European Union (EU) context. Hungary’s employment generation has been relatively strong, partly owing to the country’s favorable initial employment distribution across sectors.

Abstract

The two legs that have held up the forint in recent years—a strong “EU accession effect” and positive sentiment toward emerging markets—may no longer be strong enough to offset Hungary’s weak fundamentals. Fiscal consolidation efforts should be supported by stronger budget controls and greater transparency and accountability. This paper is an effort to shed light on Hungary’s employment dynamics, placed in the European Union (EU) context. Hungary’s employment generation has been relatively strong, partly owing to the country’s favorable initial employment distribution across sectors.

III. Employment Dynamics in the New European Member States: the Case of Hungary 58

A. Introduction

71. Low employment is a key economic and social issue in Hungary. Hungary, like most of the central and eastern European countries, experienced a sharp decline in employment in the initial years of the transition, as job shedding in government and state-owned enterprises more than offset job creation in the nascent private sector (see Schiff and others, 2006, for a description of labor market performance in transition in central and eastern European countries). A third of jobs were destroyed in Hungary from 1987 to 1995 (Kézdi, 2002). The period after the transition saw a rebound in employment, which has stabilized at a rate, low by international standards, of about 50-55 percent after 2000 (Table 1).

Table 1.

20 EU Countries: Employment Rates, 1995-2004 1/

(In percent of total population 15-64 years old)

article image
Sources: Eurostat, and IMF staff calculations.

NMS in bold.

72. This paper is an effort to shed light on Hungary’s employment dynamics, placed in the European Union (EU) context. Following Marimon and Zilibotti (1998), the evolution of employment generation is decomposed into country, industry, and temporal components. A similar decomposition is conducted for GDP growth. Comparing GDP and employment growth, the results suggest that, while convergence in the New Member States (NMS) is occurring in real economic growth terms, employment creation is lagging. The evidence for the NMS indicates that there is more variability in the evolution of employment across industries than across countries, suggesting that differences in aggregate employment rates are due to a large extent to differences in the initial employment structure than in the aggregate economic performance of the individual country. In this context, Hungary’s employment generation has been relatively strong, partly due to the country’s favorable initial employment distribution across sectors.

73. The paper also examines the determinants of a country’s employment dynamics. In this sense, the analysis makes an effort to go beyond what Marimon and Zilibotti (1998) did for the old member states of the EU. An effort is made to explain the determinants of the country-specific evolution of employment. In this context, unit labor costs per employee, real GDP, the tax wedge, and the real effective exchange rate are found to be key determinants of employment generation.

74. The paper is organized as follows. Section B provides the analytical framework. Section C describes the data used in the panel regressions. Section D reports on the results of the analysis and discusses other key determinants of employment that other studies have found to be particularly relevant for Hungary. Section E concludes.

B. Analytical Framework

75. In order to analyze employment dynamics, employment generation is decomposed into country, industry, and temporal effects. A statistical model that disentangles country-specific and industry-specific components of the generation of employment at the sectoral level is used, as in Marimon and Zilibotti (1998). Two motivations prompted the use of such a model. First, there are significant comovements of employment at the industry level across countries. These can be thought as sector (or industry) specific (e.g., worldwide sectoral technological trends) or aggregate effects (e.g., the international business cycle). Second, the generation of employment is affected by country-specific factors, such as labor costs, exchange rate movements, technology and quality upgrading of the production structure, labor legislation, and fiscal and monetary policies. Country-level effects can have either an aggregate or sector-specific nature. The model also allows to decompose employment growth into short and long term components.

76. Six different factors affecting employment are identified. The specification used is the following (see Appendix):

E(i,n,t)=H(i)+M(i,n)+B(t)+F(i,t)+G(n,t)+u(i,n,t)  ,

where E(i,n,t) represents the growth rate of employment for sector i, country n, at time t. The first factor, H(i), is a time-invariant trend component specific for each sector and shared by all countries. It represents effects such as worldwide sectoral technological trends or movements in the international price system. The second effect, M(i,n), gives the deviations between country-specific employment trends in a specific industry and the average rate across all countries for the same industry. This component represents, for example, different initial sectoral and country conditions. The third component, B(t), is a time effect shared by all countries and all sectors. It captures aggregate effects, such as the international business cycle. The fourth, F(i,t), represents industry-specific effects that cause temporary deviations from the employment trend in a specific industry in all countries at a specific time. The fifth component, G(n,t), gives country-specific aggregate effects, or the country’s transitory deviation of employment growth from the international business cycle. The last component, u(i,n,t), is a country-specific disturbance.

77. The evolution of employment in different economies is assessed against a benchmark constructed from the industry and temporal effects. The benchmark, or “virtual” employment, is created by filtering out all country-specific effects from actual employment. It is obtained by taking as initial condition the actual employment level at the beginning of the period and applying to it the growth rate formed by the sum of the average of the European employment in a specific industry, H(i), the average overall international business cycle, B(t), and the business cycle specific to that industry and common to all countries, F(i,t). Virtual employment provides a picture of what employment levels would have been observed in each country in the absence of any country-specific effect. The idea is to compare the actual employment performance with the level predicted by the respective initial employment structure if all local industries had behaved like the European average. The country-specific employment rate is the difference between the actual and virtual employment rates.

C. Data

78. The sample covers the period 1996-2004 and includes 20 countries of the EU. Employment data, from Eurostat and country yearbooks, are used, for the following 14 sectors (NACE (Classification of Economic Activities in the European Community) classification): agriculture, hunting, fishing, and forestry (A); mining and quarrying (C); manufacturing (D); electricity, gas, and water supply (E); construction (F); wholesale and retail trade, repair of motor vehicles, motorcycles, and personal and household goods (G); hotels and restaurants (H); transport, storage, and communication (I); financial intermediation (J); real estate, renting and business activities (K); public administration and defense, and compulsory social security (L); education (M); health and social work (N); and other community, social, and personal service activities (O). The countries considered are the EU countries, excluding Cyprus, Luxembourg, Malta, Portugal, and Sweden, for which data were not available.

79. Real GDP and real labor costs per employee are also analyzed.59 Sectoral data for real GDP, following the same classification described above, are from Eurostat, and for labor costs from Eurostat, country yearbooks, and the OECD structural analysis (STAN) database. Data are available for 18 countries (the 20 mentioned above, excluding Ireland and the United Kingdom). For the old member countries of the EU, the period available is 1996-2003. Other variables considered in the analysis are the tax wedge, defined as the relative tax burden for an employed person with low earnings, as published by Eurostat; the real effective exchange rate from the IMF’s Information Notice System; and the unit value ratio, calculated as the ratio of a country’s export unit values relative to the global average (58 countries, or almost 94 percent of world trade).60 The trade data for the unit value ratio are from the UN Comtrade database.

D. Explaining Employment Generation

80. Employment generation has been generally weaker in the NMS than in the old member states of the EU. As noted, the employment benchmark is the so-called “virtual” employment (obtained by filtering out the country-specific components): it is also the employment that would have prevailed if the dynamics were dictated purely by sectoral factors and shocks common to the region. Figure 1 and 2 show that the employment benchmark constructed from the group of old and new member states is higher than the one built considering only the NMS. The implication is that employment growth would have been higher in the NMS if its pace had been the same as in the old member states. At the same time, the reverse situation is shown for real GDP growth (Figure 3 and Figure 4). In this case, the benchmark is higher when only the NMS are considered. This would suggest that, while real growth in the NMS is converging to western European levels, employment is lagging.

Figure 1.
Figure 1.

NMS: Employment Growth Relative to Virtual Employment Growth, Using NMS Benchmark, 1996-2004 1/

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A003

Sources: Eurostat; country yearbooks; and IMF staff calculations.1/ Benchmark constructed from the New Member States of the EU. Cyprus and Malta are not included.
Figure 2.
Figure 2.

NMS: Employment Growth Relative to Virtual Employment Growth, Using EU Benchmark, 1996-2004 1/

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A003

Sources: Eurostat; country yearbooks; and IMF staff calculations.1/ Benchmark constructed from 20 EU countries. The 20 countries are the EU-15, excluding Luxembourg, Portugal, and Sweden, and the new EU-10, excluding Cyprus and Malta.
Figure 3.
Figure 3.

NMS: Real GDP Growth Relative to Virtual GDP Growth, Using NMS Bechmark, 1996-2004 1/

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A003

Sources: Eurostat; and IMF staff calculations.1/ Benchmark constructed from the New Member States of the EU. Cyprus and Malta are not included.
Figure 4.
Figure 4.

NMS: Real GDP Growth Relative to Virtual GDP Growth, Using EU Benchmark, 1996-2003 1/

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A003

Source: Eurostat; and IMF staff calculations.1/ Benchmark constructed from 18 EU countries. The 18 countries are the EU-15, excluding Ireland, Luxembourg, Portugal, Sweden, and the United Kingdom, and the new EU-10, excluding Cyprus and Malta.

81. Employment generation trends differ substantially across sectors. The long-term trend components of employment growth identified by the model are the yearly average European sectoral growth rates H(i) and the country-specific deviations from itM(i,n) (Table 2). Looking at the H component, employment in agriculture fell steadily, as expected, at an annual rate of 3½ percent. The decline in employment in agriculture was particularly severe during the period 2000-04 (Table 3 and 4). The other industries that appear to have expelled labor force in net terms, though at varying speeds, have been manufacturing; mining and quarrying; and electricity, gas, and water supply. The annual decline in employment in these sectors, in particular agriculture, appears to have been much more severe than if the analysis were conducted only among the NMS (Table 5). Moreover, taking into consideration only the NMS (Table 5), more industries seem to have expelled labor force than when the old member countries are also considered (Table 2). Meanwhile, real estate and business activities, hotels and restaurants, and construction appear to be the sectors where more employment has been generated. Looking at the international business cycle effects, B(t), for the restricted group of the NMS (Table 5), it can be noted that 2000 recorded the worst slump, as several NMS were going through a restructuring phase, partly triggered by the 1998 Russian crisis.

Table 2.

EU Countries: Employment Generation Components, 1996-2004

(In percent)

article image
Table 3.

20 EU Countries: Employment Generation Components, 2000-04

(In percent)

article image
Source: IMF staff calculations.
Table 4.

20 EU Countries: Employment Generation Components, 1996-99

(In percent)

article image
Source: IMF staff calculations.
Table 5.

NMS: Employment Generation Components, 1996-2004

(In percent)

article image
Source: IMF staff calculations.
Table 2.

20 EU Countries: Employment Generation Components, 1996-2004 (concluded)

(In percent)

article image
Source: IMF staff calculations.

82. Sectoral effects in the NMS account for almost two-thirds of the long-run differentials across countries and industries in employment growth; this is less the case for the large group of 20 European countries. By decomposing the variance of the time-average employmente(i,n)=1/TΣt=1TE(i,n,t) into sectoral effects, he(i), and country-specific deviations, me(i,n), we find that more than 60 percent of the total variation in long-term trends is explained by industry effects, which are country independent (Table 6). This would suggest that the initial employment structure played a key role in the uneven performance of aggregate employment rates. However, this becomes less of a factor in the larger group of European countries, where approximately half of the total variation in trends is explained by industry-specific factors and half by country-specific deviations (Table 7). Also different is the picture regarding the relevance of the components of short-term variability explained by the model for the smaller and the larger groups of countries. When only the NMS are considered, the international business cycle effects account for 8 percent of the total short-term variability (be(t)), while the temporary industry-specific effects account for 50 percent (fe(i,t)), and the country-specific effects account for 42 percent (ge(n,t)). For the larger group of EU countries, the business cycle effects appear to be even less relevant, while the country-specific effects account for almost two-thirds of the short-term variability explained by the model.

Table 6.

NMS: Analysis of Variations, 1996-2004

article image
Source: IMF staff calculations.
Table 7.

20 EU Countries: Analysis of Variations, 1996-2004

article image
Source: IMF staff calculations.

83. Among the 20 European countries, Hungary’s employment performance has been relatively good. The country-specific deviations of employment growth from the long-term sectoral trends are shown in Table 4 (matrix M(i,n)). Among the NMS, Hungary has been one of the best performers, with manufacturing, construction, and real estate and business activities generating substantially more employment than the European average. However, in the larger group of countries, Ireland and Spain have been by far the best performers, having generated more employment than average in all sectors. The Czech Republic appears to have had the worst employment generation performance. Hungary, meanwhile, compared with its virtual path by sector (Figure 5), has outperformed in all sectors in employment generation since 1998, with the exception of agriculture, in which it has performed in line with its virtual expectations.

84. Hungary’s relatively good employment generation is partly explained by its favorable initial employment distribution across sectors. A comparison of employment allocation across sectors in the NMS in 1995 shows that Hungarian employment in agriculture, the sector where the greatest decline in employment occurred, was small relative to total employment (Table 8). This favorable employment structure reflects in part Hungary’s status as one of the first economic reformers among the NMS: it began reallocating labor from agriculture to other sectors, in particular services, earlier than the other countries.

Figure 5.
Figure 5.

NMS: Employment Growth Relative to Virtual Employment Growth by Industry, Using NMS Benchmark, 1996-2004 1/

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A003

Source: IMF staff calculations.1/ Benchmarks are constructed from the New Member States of the EU, excluding Cyprus and Malta.
Table 8.

NMS: Employment Allocation, 1995

(In percent of total employment)

article image
Sources: Eurostat; and country yearbooks.

Includes wholesale and retail trade, repair of motor vehicles, motorcycles, and personal and household goods; hotels and restaurants; transport, storage, and communication; financial intermediation; real estate, renting and business activities; public administration and defense, and compulsory social security; education; health and social work; and other community, social and personal service activities.

85. The sharp decline in Hungary’s real labor costs in the second half of the 1990s and the relatively good real GDP performance may have also helped employment generation. Real labor costs appear to have been lower than the virtual real costs, in particular toward the end of the 1990s, reflecting the impact of the 1995-97 stabilization package. This was followed by a period of positive but low growth rates of real earnings (1998-2000) (Figure 6 and 7).61 The contained real labor costs, together with relatively good real GDP growth, as shown by the deviations from their virtual path (Figure 6 and 7, and Figure 3 and 4), appear to have helped employment generation.

86. Panel regression results confirm that real labor costs per employee and real GDP are key determinants of the employment rate for the NMS. Table 9 shows the results of regressing the country-specific component of the employment rate on the country-specific components of real labor costs per employee and real GDP, using a fixed-effects estimator. Both these latter variables present the expected sign and are highly significant, suggesting that lower labor costs and higher GDP growth would favor labor generation. Fitted values are presented in Table 10.

Table 9.

NMS: Explaining the Country-Specific Component of Employment, Evidence from Panel Data Regressions

(Fixed Effects)

article image
Source: IMF staff estimates.Source: IMF staff estimates.Notes: Standard errors are in parentheses

significant at 10 percent;

significant at 5 percent;

significant at 1 percent.

The regression is performed on the country-specific component of the variable.

Table 10.

NMS: Employment Generation, 1996-2004

(In percent of total population 15-64 years old)

article image
Sources: Eurostat; country yearbooks; and IMF staff calculations;Notes: (1) results from fixed effects considering real labor costs per employee, real GDP, and the tax wedge. (2) Results from fixed effects considering real labor costs per employee, real GDP, the tax wedge, and the real effective exchange rate. (3) Results from fixed effects considering real labor costs per employee, real GDP, the tax wedge, and the unit value ratio.
Figure 6.
Figure 6.

NMS: Growth of Real Labor Costs per Employee Relative to Virtual Growth of Real Labor Costs per Employee, Using NMS Benchmark, 1996-2004 1/

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A003

Sources: Eurostat; country yearbooks; and IMF staff calculations.1/ Benchmark constructed from the New Member States of the EU. Cyprus and Malta are not included.
Figure 7.
Figure 7.

NMS: Growth of Real Labor Costs per Employee Relative to Virtual Growth of Real Labor Costs per Employee, Using EU Benchmark, 1996-2003 1/

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A003

Sources: Eurostat; OECD STAN database; country yearbooks; and IMF staff calculations.1/ Benchmark constructed from 18 EU countries. The 18 countries are the EU-15, excluding Ireland, Luxembourg, Portugal, Sweden, and the United Kingdom, and the new EU-10, excluding Cyprus and Malta.

87. The tax wedge and the real effective exchange rate are also important determinants of the employment rate. The coefficient of the tax wedge is negative and statistically significant and appears to be robust across specifications (Table 9), suggesting that lowering the tax wedge favors employment generation. This is particularly relevant for Hungary, which, although it has decreased, has one of the widest tax wedges among the NMS (Table 11). Lowering the tax wedge should reduce employee costs and help generate employment. The real effective exchange rate is also a relevant determinant (Table 9), suggesting that a depreciation of the real effective exchange rate would improve competitiveness and help employment creation.

Table 11.

NMS: Tax Wedge, 1996-2004 1/

(In percent)

article image
Source: Eurostat.

The tax wedge on the labor cost measures the relative tax burden for an employed person with low earnings.

88. Although it presents the expected sign, technological upgrading does not appear to be statistically significant. Higher product technology (or higher product quality), proxied by the unit value ratio presents the right sign, suggesting that upgrading the production structure with higher technology would reduce employment generation. However, trade specialization does not appear to be statistically significant. In the context of Hungary, this would be in line with the firm-level results of Köllő (2006), who shows that there is no straightforward linkage between product quality and the skill composition of the labor force manufacturing the product. In particular, he shows that the labor demand effects of trade specialization in the late stage of the transition were not necessarily detrimental to low-skilled labor, which is the group that would be expected to have suffered most from technological upgrading. Since our sample period covers mostly the late stage of the transition, our results would not contradict the presumption that the rapid shift of the Hungarian output structure to relatively skill-intensive activities lowered the demand for unskilled workers, at least in the initial phase of the process.

89. The robustness of these results is confirmed when the analysis is performed on country-specific employment growth at the industry level; however, some differences across sectors emerge. Regressions for the 14 NACE sectors of the eight NMS, for the period 1997-2004, confirm that the country-specific growth rate of the labor costs and of real GDP, the changes in the tax wedge, and the real effective exchange rate are relevant for explaining the country-component of employment generation62 (Table 12). However, changes in real labor costs and real GDP growth appear to have the largest effects on sectors such as construction, business activities, and health.63 The tax wedge, meanwhile, seems equally relevant for all sectors.

Table 12.

NMS: Explaining the Country-Specific Component of Employment at Sectoral Level, Evidence from Panel Data Regressions (Fixed Effects) 1/

article image
Source: IMF staff estimates.Notes: Standard errors are in parentheses

significant at 10 percent;

significant at 5 percent;

significant at 1 percent.

Sectors A-O defined in para. 6.

The regression is performed on the country-specific component of the variable.

90. Other researchers have found education to be an important factor for employment generation. As observed by Commander, K∆llő, and Tolstopiatenko (2004), while in general minimally educated workers flow in and out of low-wage unstable jobs with relative short periods of unemployment between jobs, in central Europe the large majority of the minimally educated remains unemployed for protracted periods. This phenomenon appears particularly severe in Hungary, where the employment ratio is very low among workers with only primary education (Figure 8).

Figure 8.
Figure 8.

Hungary: Employment-Population Ratios by Education, 2002

(In percent)

Citation: IMF Staff Country Reports 2006, 367; 10.5089/9781451818048.002.A003

Source: OECD (2005b).

91. Similarly, structural factors and labor institutions have also been found to be key determinants of employment. Focusing on faster-growing sectors, such as services, for the OECD countries, Messina (2005) finds evidence that laws and institutions, such as product and labor market regulations, hamper the expansion of services employment. Pierre and Scarpetta (2006) show that strict employment protection regulations can have negative effects on job creation because they weaken a firm’s ability to take advantage of the opportunities offered by new technologies and the access to new markets, which often require a change in the skill composition of the workforce. While Hungary does not stand out for the strictness of its labor regulations, at least among the NMS (Table 13), the welfare system—in particular, the disability pension—is quite generous (Table 14), and both men and women have dropped out of the labor force to take advantage of the system (Figure 9). The number of benefit recipients has grown massively over the past decade, as the number of disability benefits awarded to those below the standard age of retirement has increased to over 10 percent of the working-age population since the early 1990s (OECD, 2005a). In this context, Cseres-Gergely (2006), based on a model of retirement decision, presents evidence that incentives provided by the pension system, especially the disability pension, make retirement very attractive, thereby reducing the labor supply.

Table 13.

NMS: Indicators of the Strictness of Employment Regulation, 2005

(Scale: 0-100, with 100 the most restrictive)

article image
Source:http://rru.worldbank.org/DoingBusiness/ExploreTopics/HiringFiringWorkers.