V. Potential Output growth in Hungary42
125. Hungary is one of the most advanced transition economies. Having started its transition already in the early 1980s, by 1990 as much as 75 percent of Hungarian prices were already market determined and the private sector was expanding rapidly. Yet since the late 1980s, Hungary’s output growth performance has been relatively modest compared to neighboring transition economies. By 1997, official data indicate that real GDP in Hungary was still around 9 percent below its level in 1988; this contrasts with the Czech Republic and Poland where real GDP in 1997 were equal to or exceeded the 1988 levels by 12 percent, respectively.
126. This Chapter employs a neoclassical growth accounting framework in an attempt to determine the underlying forces driving Hungary’s growth performance. This assessment is intended to form the basis for establishing a view on future potential output growth, although it can provide also some tentative indications on the current level of the output gap. In order to achieve a good understanding of the underlying growth factors, it is particularly important to understand the sharp contraction in the early 1990s and the subsequent relatively slow recovery: was economic activity suffering from insufficient demand or were other factors at play?
127. The analysis suggest that most of the collapse in output in the early 1990s was probably due to a loss of productive potential as the barter trade arrangement collapsed. The initial output drop was followed by several years of below potential growth, resulting in a widening output gap. Since last year, however, real GDP growth has picked-up thus narrowing the output gap significantly. An illustrative scenario focuses on a set of conditions (in terms of employment, investment, and total factor productivity growth) that would allow potential output to grow by 5–5¼ percent on average during 1998–2003. This scenario is in line with the results in earlier studies, for example, van Elkan (1997), Borensztein and others (1991), and EBRD (1997) which estimated that—assuming successful structural reform—that potential growth rate was in the 5–5½ percent range.
A. The Growth Accounting Framework
128. The traditional growth accounting framework dates back to Solow (1957). In this framework, GDP and GDP growth are split into the contribution from two factors of production—capital and labor—and a residual, the “Solow residual” or “total factor productivity” (TFP). It can be shown that, assuming a Cobb-Douglas production function, the natural logarithm of real GDP can be decomposed into a weighted sum of the two factors of production—where the weights are the respective factor income shares—and a residual (TFP):
where α is the capital income share. Equation 1) can also be expressed in terms of growth rates since the first difference of the growth rates of the natural logarithm is approximately equal to the percent change in the underlying variable. Furthermore, it can be shown that total factor productivity growth is equal to the sum of capital productivity growth (output growth minus percent change in the capital stock) and labor productivity growth weighted by income shares. The growth of “labor” can be broken down into change in the working-age population, change in the trend participation rate, and change in the natural rate of unemployment.
129. The basic growth accounting framework can be used to assess the level and growth of potential output. This is done by distinguishing between actual employment (L) and potential employment (L*). Potential employment is usually derived from an estimate of the “non-accelerating inflation rate of unemployment” (NAIRU=u), structural labor market participation rate (Pa), and the working-age population (Pw):
The capital stock is generally assumed to be fully utilized, but a “capacity utilization gap” can also be proxied by using survey data. Potential TFP growth is normally estimated by applying a Hodrick-Prescott or other filtering techniques to the natural logarithm of the difference between real output and the weighted product of potential employment and the capital stock.
130. In addition to the standard qualifications to this approach (in particular, the assumed production function), several issues are paramount when applied in a transition context. First, capital and labor were probably underutilized in the era of state planning. Second, the transition period led to a massive shift of demand, rendering a large proportion of the existing capital stock obsolete. Third, official figures probably overestimate the fall in real output from the planning period—when barter, quotas, and price regulations led to substantial economic distortions—to the transition years when a whole new range of better quality products became available.
131. In order to use the growth accounting framework to assess the future growth potential of the Hungarian economy, it is first necessary to estimate the current level of capital. This is required since the growth impact from a given investment–to–output ratio depends on the size of the initial capital. In the next section, the growth accounting framework is applied to Hungary’s recent history, tracking the period from 1985, before the transition started, until 1997. This will provide an estimate of historic and actual potential output, the output gap, and the capital-output ratio from 1985 to 1997.
B. Potential Output: 1985–97
132. Capital stock data for 1985–90 were based on official data on the nominal value of the capital stock (Central Statistical Office, 1993). In order to transform the series into fixed 1995-price values, the nominal capital stock values were deflated by using the fixed investment deflator from national accounts. In spite of the likely poor quality of the data, the pre-transition capital stock data provides a valuable starting point for the estimates of post-transition capital stock.
133. The sharp contraction triggered by the collapse of CMEA trade in 1990–91 overshadows smaller cyclical fluctuations which are the main components of the output gap in mature market economies. The cumulative loss in GDP from 1990 to 1993 exceeded 18 percent. As a result of this major structural break, standard methods used to trace potential output, such as a detrending of real GDP or the use of Vector Autoregressions (VAR), are not applicable to Hungary or other transition economies43. These methods are based on the assumption that real GDP can be described as the outcome of a “smooth” stochastic process, and so are not applicable to the context of large one-off “discrete” structural shocks.
134. Thus, in order to avoid the problem with these time series methods, a one-off reduction in the capital stock was assumed in 1990 and 1991, the time of the collapse of the CMEA trade (Figure 16). An approximation of the warranted reduction was based on estimates of the direct and indirect effects of the trade shock and the effect of “disorganization” (Box 1). A one-off cut of 35 percent was introduced in 1991, corresponding to some 20 percent loss due to the collapse of CMEA trade and 15 percent due to “disorganization.”
Figure 16.Hungary: Components of Potential Output, 1985–1997
Sources: Central Statistical Office and staff estimates.
Box 1.The Loss of Productive Potential
1. The breakdown of CMEA trade exposed Hungary to a combination of a substantial demand shock and a terms of trade shock. The collapse in CMEA import demand dramatically reduced Hungarian export volumes: from 1989 to 1992, Hungarian exports to the CMEA area fell by some 50 percent in real terms. At the same time, the collapse of barter trade led to a large terms of trade deterioration, and a sharp rise in the oil price. The terms of trade loss was compound by a sharp rise in the international oil market price as a result of the Gulf war.
2. The effect of the drop in external demand induced a loss in the productive potential of the economy as it rendered obsolete capital in industries that relied on CMEA exports. In most cases, the redirection of exports to Western markets meant a disruption of previous production lines. As a result, capital equipment used to produce the lower grade products traded within CMEA (food products, textiles and clothing, and various machinery and appliances) had to be scrapped as it was inadequate to produce the higher quality products demanded by developed market economies. Furthermore, embedded human capital, such as product specific know-how and informal networks, was disrupted as a result of the dramatic change in demand. For a theoretical and empirical discussion of the costs of “economic disorganization,” see Blanchard (1997).
3. A further loss in productive potential arose from trade liberalization, which was significantly accelerated in 1990–91. The entry of new goods lowered the market clearing price of domestically produced goods to below its short-run break-even level, disrupting the production of these goods and scrapping product-specific real and human capital. The scrapping process was accelerated by the large increase in energy costs. In addition, the bankruptcy legislation of 1992, while having overall positive effects, resulted in the liquidation of a number of economically viable companies.
4. The loss in productive potential arising from these factors is difficult to measure. A lower bound estimate can be obtained by focusing on the direct and indirect effects of the collapse in CMEA exports. According to the OECD (1993), about one-third of the fall in industrial output from 1988 to 1991 can be ascribed to direct and indirect effects of the collapse in ruble exports, using input-output tables at detailed sector level. With a drop in industrial output of about 30 percent in the period (of which 9½ percent was directly or indirectly due to the drop in rouble exports), this alone would have contributed to a 7 percent drop in potential output (given an output share of manufacturing industries of about 70 percent). Under the assumption of a Cobb-Douglas function and a capital income share of about 40 percent, such an output drop corresponds to a 20 percent fall in the capital stock. The other effects discussed above come on the top of the effect of the fall in export demand.
135. Capital accumulation in the post-transition period was based on data on new investment in machinery and equipment and construction, and the assumption of an annual depreciation rate of 20 percent for machinery and equipment and 5 percent for real estate (corresponding to an average life of 5 years and 20 years, respectively). These depreciation rates are in line with international standards, for example, as used by the OECD. No correction was made for the potential larger productivity of the new investment, as it was assumed that, partly, the old capital equipment would be reallocated and that the full potential of the new investment would only be reached gradually and would be appropriately included into TFP growth44.
136. Developments in potential labor supply are equally difficult to determine since the unemployment rate surged and the participation rate dropped sharply in the wake of the transition (Figure 17). In such an environment, estimates of the “natural” rate of unemployment or the NAIRU are highly uncertain. One key question is whether the workers fired from previously state-owned companies were employable in the new private industries. While the generally high level of education of Hungarian workers suggests that these workers are relatively easily employable, the market economy requires different skills, which may be difficult to acquire for the older workers accustomed to central planning and state ownership. In the absence of reliable microeconomic data on the extent of skill-mismatch, the concomitant increase in unemployment and vacancies during 1991–93 suggest that a significant share of the increase in unemployment has been structural.45 Consequently, the share of the long-term unemployed has also increased rapidly, albeit still significantly less than in the average of the EU. Given these uncertainties, two alternative estimates of potential labor supply are presented: one is based on a “low” NAIRU, where the NAIRU is assumed equal to a moving average of long-term unemployment; the other, the “high” NAIRU estimate, was based on a simple average between long-term unemployment and actual unemployment (Figures 16 and 17).
Figure 17.Hungary: Components of Potential Employment, 1985–1997
Sources: Central Statistical Office and staff estimates.
137. The contribution of labor and real capital was added together by using an average capital income share as reported by the Hungarian national accounts in 1993–97 which, at 39 percent, is significantly above the international average (25–30 percent).46 The capital income share is also large in relation to Hungary’s neighboring transition economies and reflects moderate wage developments, labor shedding, and the success of structural reform in redressing the return on fixed investment. TFP or the “Solow residual” was estimated by smoothing the natural logarithm difference between actual output and the weighted product of the capital stock and potential employment.
138. Adding up the natural logarithm of all the components—the weighted sum of the input factors potential employment and capital, explained TFP, and the smoothed residual—provides an estimate of the level of the natural logarithm of real potential output. The resulting estimate of potential output indicates that the collapse of output in 1989 to 1992 and the subsequent slow growth were mainly a result of a drop in the productive potential of the economy (Figure 18). Indeed, actual output may have exceeded potential output in 1991 and 1992, Since 1994, however, both in the case of the low and high NAIRUs the estimates suggest that potential output has outpaced actual output by some margin, resulting in a negative output gap.47
Figure 18.Hungary: Actual and Potential GDP, 1985–1997
Sources: Central Statistical Office and staff estimates
C. Prospects for Future Growth
139. The previous section established a starting point for assessing the conditions, in terms of employment, investment and total factor productivity growth, that would allow Hungary to grow at a targeted growth rate. For illustrative purposes, we focus on a possible GDP growth target of 5–5¼ percent over the medium-term. The scenario (Table 7) shows that growth rates in this range can be achieved with a relatively contained increase in the investment-to-GDP ratio. The scenario is based on the following assumptions:
|Potential output (A+B+C)||3.2||4.1||4.8||5.0||5.3||5.2||5.1|
gross fixed investment grows on average by 8 percent per year, leading to an increase in the fixed investment-to-GDP ratio from 23½ percent in 1998 to 27 percent in 2002;
the NAIRU gradually declines from 6½ percent in 1998 (the average of the high NAIRU and low NAIRU scenarios discussed in the previous sections) to 5 percent in 2002, as a result of labor market reform, particularly cuts in the tax wedge;
the participation rate gradually increases from 61 percent in 1997 to 70 percent in 2002, close to its pre-transition level; this should again, reflect the effect of labor market reform;
the working-age population shrinks fast (¾ percent per year), in line with World Bank projections, reflecting the retirement of the large postwar “baby boom” generation. This is expected to subtract almost ½ percent per year from the potential growth rate;
total factor productivity increases by 2½ percent per year; this is in line with the experience of many emerging market economies: during 1978–96, for example, TFP in Thailand and Singapore is estimated to have increased by 2 percent and 2¼ percent respectively (Sarel, 1997). However, this is well below the post-war experience of larger European countries and Japan: during 1950–73, TFP in Japan and the three largest EU countries (Germany, France, and Italy) is estimated to have increased by 3½ percent and 3¼ percent per year, respectively.
It must be stressed that different assumptions regarding investment have a relatively small effect on the growth rate of potential output. This reflects the relatively high capital-output ratio, even after the large negative shocks introduced in 1990–91. By contrast, different assumptions regarding the trend in the NAIRU or the participation rate have more powerful effects for the period considered. This underscores the importance of maintaining a high degree of flexibility in labor and product markets and of lowering labor taxation.
Barro, Robert J., and XavierSala-I-Martin, 1995, Economic Growth (New York: McGraw-Hill).
Blanchard, Olivier, and MichaelKremer, 1997, “Disorganization,”mimeo (Unpublished: MIT).
Borensztein, Eduardo, and PeterMontiel, 1991, “Saving, Investment, and Growth in Eastern Europe,” IMF Working Paper 91/61 (Washington: International Monetary Fund).
Burnieaux, Jean-Marc, 1995, “Establishing Financial Discipline: Experience with Bankruptcy Legislation in Central and Eastern European Countries,”OECD Economic Studies, No. 25 (Paris: OECD).
Central Statistical Office,1993, Statistical Yearbook of Hungary, Budapest.
Christoffersen, Peter, and PeterDoyle, 1998, “From Inflation to Growth: Eight Years of Transition,” IMF Working Paper 98/100 (Washington: International Monetary Fund).
de Melo, Martha, CevdetDenizer, and AlanGeld, 1997, “from Plan to Market: Patterns of Transition,”inMario J.Blejer and MarkoŠkreb (eds.), Macroeconomic Stabilization in Transition Economies,Cambridge University Press, Cambridge, United Kingdom.
European Bank for Reconstruction and Development,Transition Report 1997, (London).
Fisher, Stanley, RatnaSahay, and CarlosVégh, 1996, “Stabilization and Growth in Transition Economies: The Early Experience,” IMF Working Paper 96/31 (Washington: International Monetary Fund).
Harberger, Arnold C., 1998, “A Vision of the Growth Process,” American Economic Review, Volume 88, No. 1.
Hviding, Ketil, 1998, “Switzerland’s Long–Run Growth Slowdown,”inSwitzerland—Selected Issues and Statistical Appendix, SM/98/40.
Magnier, Antoine, 1998, “Potential Output and the Output gap in Austria: A European Perspective,”Austria—Selected Issues and Statistical Appendix, SM/98/126 (Washington: International Monetary Fund).
Mizsei, Kalman, 1993, “Bankruptcy and the Postcommunist Economies of East-Central Europe,” Russian and East European Finance and Trade, No. 30, Pp. 34–70.
Sarel, Michael, 1997, “Growth and Productivity in ASEAN Countries,” IMF Working Paper 97/97 (Washington: International Monetary Fund).
Solow, Robert, 1956, “A Contribution to the Theory of Economic Growth,” Quarterly Journal of Economics; pp. 65–94, February.
Prepared by Ketil Hviding
See Magnier (1998) for the application of several different time-series techniques to estimate potential output and the output gap in Austria.
See Borensztein and others (1991) for an estimation of the relative productivity of pre-transition fixed investment in Hungary, Poland, and Czechoslovakia.
For details see OECD Economic Survey of Hungary, 1997 (p. 75).
The capital income share was calculated as the ratio of net property income to nominal GDP.
The output gap is estimated to be some 1½–2 percent of GDP in 1997. However, owing to the high degree of dependency of this result on the assumption made on potential labor supply and TFP, this estimate should be treated with caution.