WPS4384 Policy ReseaRch WoRking PaPeR 4384 More Growth or Fewer Collapses? A New Look at Long Run Growth in Sub-Saharan Africa Jorge Saba Arbache John Page The World Bank Africa Region Office of the Chief Economist November 2007 Policy ReseaRch WoRking PaPeR 4384 Abstract Low and highly volatile growth define Africa's growth collapses, offsetting most of the benefits of growth. Had experience. But there is no evidence that growth volatility Africa avoided its growth collapses, it would have grown is associated to long term economic performance. This 1.7 percent a year instead of 0.7 percent, and its GDP result may be misleading if it suggests that volatility is not per capita would have been more than 30 percent higher important for economic and social progress. In this paper in 2005. The authors also find that growth accelerations we use a variant of the method developed by Hausmann, and decelerations have an asymmetric impact on human Pritchett, and Rodrik (2005) to identify both growth development outcomes. Finally, our results suggest acceleration and deceleration episodes in Africa between that it is easier to identify the likely institutional and 1975 and 2005. The authors find that Africa has had policy origins of growth decelerations than of growth numerous growth acceleration episodes in the last 30 accelerations. years, but also nearly a comparable number of growth This paper--a product of the Office of the Chief Economist , Africa Region--is part of a larger effort in the department to investigate growth and poverty in Africa. Policy Research Working Papers are also posted on the Web at http://econ. worldbank.org. The author may be contacted a tjarbache@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team More Growth or Fewer Collapses? A New Look at Long Run Growth in Sub-Saharan Africa Jorge Saba Arbache and John Page The World Bank Keywords: Growth acceleration and deceleration, Sub-Saharan Africa. JEL Code: O11, O47, O55, O57. Acknowledgements: We would like to thank Vijdan Korman for excellent research assistance, and Monica Das Gupta for comments and suggestions. Disclaimer: The findings and interpretations of this paper are those of the authors. They do not represent the views of the World Bank, its Executive Directors or the countries they represent. Corresponding author: Jorge Saba Arbache, World Bank, 1800 H St, NW, Washington, DC, 20433; email jarbache@worldbank.org. 1. Introduction During the last three decades growth in Sub-Saharan Africa (hereafter Africa) has been both low and highly variable (Ndulu et al., 2007). Between 1975 and 2005 per capita income PPP grew by 0.7% per year, by far the lowest figure among developing regions. At the same time country growth rates were highly volatile. Interestingly, however, there is no evidence that growth volatility was associated with Africa's poor long term economic performance (Arbache and Page, 2007a). This result is unexpected (Ramey and Ramey, 1995, Hnatkovska and Loayza, 2004) and may be misleading. Perhaps because no statistical association exists between Africa's long term growth rate and its volatility, most attempts to explain Africa's growth performance have focused on investigating the determinants of growth overtime and across countries using standard models and techniques (Ndulu et al., 2007, O'Connell and Ndulu, 2000, Collier and Gunning, 1999). Instead, given Africa's high growth volatity, it may be more relevant and rewarding to examine the causes and consequences of medium term deviations from the long run trend ­ growth accelerations and decelerations. Building on the work by Hausmann, Pritchett and Rodrik (2005) (hereafter HPR), we propose in this paper an empirical method to identify growth accelerations and decelerations relative to a country's long term growth trend. Focusing on both good times and bad times gives us a broader view of the growth experience, which we believe to be particularly relevant for Africa. To reflect the heterogeneity of African countries' long run performance, we endogenize economic conditions into the methodology by defining episodes of growth and decline relative to each country's long run trend growth. In contrast, HPR and related literature impose common parameters across countries to identify growth accelerations. Using our methodology, we find that African countries have experienced numerous growth acceleration episodes in the last 30 years, but also a comparable number of growth collapses. In short, Africa's long run record of slow and volatile growth reflects a pattern of alternating, identifiable accelerations and declines, rather than random variations of growth rates around the long run trend. We also find that growth volatility ­ when viewed as the product of accelerations and declines ­ is not neutral and indeed matters for economic and social outcomes. To begin to address the public policy questions posed by these results, we 2 look for correlates associated with acceleration and deceleration episodes and examine the probability that an economy will undergo a growth acceleration or deceleration. This paper is organized as follows. Section 2 briefly discusses the Africa's growth experience. Section 3 presents the methodology for identifying growth accelerations and decelerations. Section 4 presents the main results. Section 5 examines whether growth accelerations and decelerations matter for economic and social outcomes and looks at correlates of accelerations and decelerations. Section 6 concludes. 2. Africa's growth 1975-2005 Data on GDP per capita (PPP at 2000 international prices) and its growth rate are taken from the World Development Indicators, unless otherwise specified. Our sample includes all Sub-Saharan countries, except Liberia and Somalia, for which there are no GDP per capita PPP data.1 Because we are primarily interested in examining the representative country, we use unweighted country data in the aggregate analysis unless otherwise stated. Our time series spans from 1975 to 2005.2 We thus have an unbalanced panel of data with T=31 and N=45. This period follows the first oil-shock and includes the commodities prices plunge, when many African economies collapsed and several conflicts erupted, the introduction of structural reforms, which brought significant changes in many economies, and the recently observed growth recovery. Figure 1 shows that mean GDP per capita in Africa had a slow, positive long term trend, consisting of about 20 years of virtual stagnation with a point of inflexion upwards in mid-1990s. Since then, actual income has remained above the trend most of the time and the variance appears to have declined.3 Figure 2 shows Africa's growth path over the same period. Trend growth declined until the late 1980s, and increased thereafter, although there is evidence of a slowdown in the last years. Variance has declined since the mid-1990s and actual growth has tended to be above and closer to the trend. 1Our sample accounts for more than 98% of population and 99% of the regional GDP in 2005. 2The WDI's GDP per capita PPP series starts in 1975. 3 We employ the Hodrick-Prescott filter in Figures 1 and 2 to smooth the estimate of the long term trend component of the GDP series. 3 Figure 1: GDP per capita 2800 2400 100 2000 0 -100 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 GDP per capita Trend Cycle Figure 2: GDP per capita growth 8 4 0 8 -4 4 0 -4 -8 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 GDP per capita growth Trend Cycle At the aggregate level, growth increased substantially during 1995-2005 and was accompanied by a sharp reduction in the coefficient of variation. The per capita growth rate rose to 1.88% in this period from -0.23% of the previous decade, and the coefficient of variation fell to 3.2% from -25.5%. This shift implies an increase of 2% in growth, which is about three-times the long term growth rate of 0.7%. Income per capita went up to $2,486 in 1995-2005, which represents an increase of about $300, or 11%, as compared to previous periods. Recursive residual estimations, Chow breakpoint tests, and Chow forecast tests, do not reject the hypothesis that a structural break in the growth series occurred between 1995 and 1997 (Arbache and Page, 2007b). 4 3. Identifying good times and bad We define a growth acceleration as a period that satisfies the following four conditions: Condition 1 ­ The forward four-year moving average growth minus the backward four-year moving average growth > 0 for a given year; i.e., the forward moving average window (t, t+1, t+2, t+3) must be higher than the backward window (t, t-1, t-2, t-3) and above 0. A signal change from (+) to (­) or vice-versa suggests growth trend shift;4 Condition 2 ­ The forward four-year moving average growth exceeds the country's average growth, meaning that the pace of growth during acceleration is higher than the country's trend; Condition 3 ­ The forward four-year moving average GDP per capita exceeds the backward four-year moving average; Condition 4 ­ A growth acceleration episode requires at least three years in a row satisfying conditions 1-3. An episode includes the three subsequent years after the last year that satisfies conditions 1-3; i.e., we attach the moving average window to the years identifying the growth acceleration.5 Condition 2 endogenizes the country's economic conditions, because its growth trend is a key parameter for identifying growth acceleration episodes. There is clearly a risk that by identifying a period of modest, sustained growth in a low growth economy as a growth acceleration episode we will assign too much significance to a minor change in economic performance. But it is also true that a period of relatively modest per capita growth, say 2%, may well be a genuine growth boom in a country enduring very low growth rates, and a decline in per capita income of equal magnitude could spell a serious economic collapse in a stagnant economy. Condition 2 also helps to limit the number of identified accelerations in countries with sustained, long run growth: if a country, for example, is growing rapidly it will lift the growth trend, reducing the number of estimated accelerations. This is particularly significant for countries experiencing very low or very high growth rates. Condition 3 ensures that the growth acceleration episode is not a recovery from a recession. 4The window size may change according to the long term growth volatility of a set of countries and/or region: the higher the volatility, the lower should be the window size if one wants to observe the effects of volatility on economic performance. 5As an example, if conditions 1 to 3 identify growth acceleration during, say, 1991 to 1995, the years 1996, 1997 and 1998 are included as part of the episode. Thus, this growth acceleration episode comprises a period that starts in 1991 and ends in 1998. 5 The identification of growth deceleration episodes requires the following adjustments: in Condition 1, the forward four-year moving average growth minus the backward four-year moving average growth < 0 for a given year; in Condition 2, the forward four-year moving average growth is below the country's average growth; in Condition 3, the forward four-year moving average GDP per capita is below the backward four-year moving average. The methodology has three main characteristics that affect the interpretation of the results: first, it identifies good and bad times rather than only focusing on rapid growth spells or deep collapses; second, it carries over inertia when identifying growth accelerations and decelerations; and third, it is sensitive to the length of time series. Figures 3-5 illustrate the methodology at work. Condition 1 alone would identify 1993-2005 in Tanzania as a growth acceleration episode (Figure 3). But only the period 1998-2005 satisfies Conditions 1-4 and is identified as a growth acceleration. Condition 3 identifies 1993-1998 is a recovery from a recession. Figure 3 : Tanzania - GDP per capita growth rate and growth acceleration 6 5 4 Conditions 1 to 4 3 satisfied (1998-05) Only condition 1 2 satisfied 1 (1993-05) 0 -1 9891 9901 9911 9921 9931 9941 9951 9961 9971 9981 9991 0002 0012 0022 0032 0042 0052 -2 -3 -4 The cases of Senegal and South Africa illustrate the method further. The dotted lines show deceleration periods, while the shaded lines identify acceleration periods. Senegal experienced a contraction between 1988 and 1994; the average growth rate during this period 6 was -1.4%, well below the average of 0.35% (Figure 4). In contrast, average growth in 1994- 2001 ­ a growth acceleration ­ was 1.75%. Figure 4: Senegal - GDP per capita growth rate and growth acceleration and deceleration 15 10 Growth acceleration 5 (1994-01) 0 5 7 9 1 3 5 7 9 1 3 5 7 9 1 3 5 197 197 197 198 198 198 198 198 199 199 199 199 199 200 200 200 -5 Growth deceleration (1988-94) -10 In the case of South Africa, the average growth rates during the downturns 1982-1987 and 1989-1994 were -1.91% and -1.54%, respectively, compared to the overall mean of 0.12% (Figure 5). During the growth acceleration period, 1999-2005, growth was 1.96%. Table A1 in the Appendix shows the start years of growth acceleration and deceleration episodes and compares our results with those from two closely related studies. HPR in their seminal work identify growth accelerations worldwide using GDP per capita growth from 1950 to 1999 from the Penn World Tables. They compare seven-year forward and backward moving average growth windows and impose a cutoff that the forward moving average window should exceed the backward moving average window by at least 2%. They further require that average growth must be at least 3.5% during the acceleration episode. 7 Figure 5: South Africa - GDP per capita growth rate and growth acceleration and deceleration 5 4 Growth acceleration 3 (1999-05) 2 1 0 75 77 79 81 83 85 87 89 91 93 95 97 99 01 03 05 -1 19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20 -2 -3 Growth deceleration (1989-94) -4 Growth deceleration (1982-87) -5 Pattillo, Gupta, and Carey (2005) (hereafter PGC) have applied a similar methodology to Africa only, using GDP per capita PPP growth between 1980 and 2004 from the IMF World Economic Outlook.6 To address the higher volatility and lower overall rate of growth in Africa the authors use a five-year window and require that growth average be at least 2% during the acceleration. The main differences between these two methodologies and the one proposed here are that: first, the moving average windows are bigger than ours. Second, they impose a cutoff of at least 2% in the forward minus backward moving average window, whereas we impose a cutoff of zero. Third, they impose a common minimum growth rate to define an acceleration for all countries, whereas we use the country's growth trend as the cutoff. In general, our calculations accord with those the other two studies, but since our filter is more flexible ­ and identifies decelerations ­ it picks up more episodes; we find 32% more episodes for Africa than PGC and 114% more than HPR. 6PGC calculated the growth acceleration episodes using the HPR methodology and PPP growth data. So results in Table A1 are fairly comparable to ours, despite slight differences between the IMF and WDI data. The IMF's and WDI's data generally follow the same pattern. However, the WDI's GDP per capita tends to be slightly lower than the IMF's. For a discussion on the discrepancies between the IMF and WDI GDP data see Africa Development Indicators 2006 (p. 114). 8 4. Results Table 1 shows the frequency of accelerations and decelerations and their associated growth rates during selected periods. For the full period there is a slightly higher probability of a growth acceleration than deceleration: 25% of the 1,243 country-year observations (total of valid observations per country per year) belong to growth accelerations, while 22% are classified as growth decelerations.7 Between 1975 and 2005 countries in Africa that experienced growth accelerations managed to grow on average by 3.6% during those episodes, compared with the region-wide average of 0.7%. During decelerations countries contracted on average by -2.7%. Given the almost equal probabilities of growth accelerations and decelerations, most of the benefits of growth accelerations in the continent were offset by growth collapses, leading to the region's overall tepid rate of growth. Had Africa avoided its growth collapses it would have grown at 1.7% a year in per capita terms instead of 0.7%. Figure 6 shows the actual and simulated GDP per capita at these growth rates. Income per capita would have been at least 30% higher in 2005 from avoiding bad times.8 Growth decelerations matter a great deal for fighting poverty in Africa The relative frequency of good and bad times is reflected in Africa's long run pattern of growth. Accelerations are more frequent in 1995-2005; decelerations are more common in the two preceding decades. Forty two percent of the 494 country-year observations of 1995- 2005 occur in countries experiencing growth accelerations, and only 12% in countries undergoing growth decelerations. The remaining 46% of observations belong to years in which countries were experiencing neither growth acceleration nor deceleration. In 1975- 1984 growth decelerations were 350% more frequent than accelerations.9 In 1985-1994 this ratio had dropped to 71%, mainly due to a sharp rise of accelerations to 21% from 4%. 7To check the robustness of our results we also identified growth accelerations and decelerations by replacing 0 with +1% and -1% for acceleration and deceleration, respectively, in condition 1, but the results did not change substantially. We therefore report only the base case results because they are less restrictive. 8 The simulated growth rate without collapses takes into account growth rate during all country-years but growth deceleration years. The additional GDP per capita results from the difference in compound growth at 1.7% and 0.7% in 1975-2005. 9Calculated as ((0.18/0.04)-1)*100. 9 per 4 5 4 7 4 PDG 10 capita 1,79 1,76 ,801 1,79 ,791 eleration rate 4 6 dec th ht owrG 2.7- 3.0- 8.13- 9 1.2- 4.13- capitar rowG ncy pe try-n )s PDG year .220 .180 0.36 .120 0.29 Freque (cou and, per 8 6 7 9 2 rates th PDG capita 2,59 2,79 ,902 2,44 ,892 on grow tion, eratile rate acc th 4 1 6 th decelera owrG owrG 3.6 4.6 1.23 3.7 9.33 and ncy try-n )sr 5 4 yea 0.2 0.0 1.20 2 0.4 4.10 Freque (cou acceleration th per 9 0 3 6 2 grow of PDG tai cap ,292 ,182 2,18 ,482 2,18 PDG th equency ow 3 7 Fr gr 1: bleaT PDG 0.70 3.10 0.2- 8.81 0.0- 3 6 3 4 9 ervations 1,24 31 43 49 74 Obs 05 84 94 05 94 20 19 19 20 19 Period 1975- 1975- 1985- 1995- 1975- Figure 6: Actual and simulated GDP per capita 3,600 3,400 964, 3,200 $3 3,000 2,800 936, $2 2,600 9 2,400 2 57 09 $2, 2,200 $2, 2,000 75 77 79 81 83 85 87 89 91 93 95 97 99 01 03 05 19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20 Actual GDP per capita GDP per capita growth at the observed average (0.7% pa) GDP per capita growth in the no-collapse scenario (1.7%pa) In 1995-2005, the average growth rate for countries during acceleration episodes was 3.8%, the second highest average among the three ten year periods. Interestingly, it was in 1975-1984, a period of very modest regional economic growth, that average growth during accelerations reached its highest rate. This reflects a compositional effect at work. In the last decade even long stagnant economies such as the Central African Republic, Ethiopia, Mali, Mozambique, Sierra Leone, and Tanzania, experienced some sustained growth, pushing down the averages during acceleration episodes, whereas in 1975-1984 the high average growth rate was mainly due to a small number of growth accelerations over all and very rapid growth in the Republic of Congo. The average (negative) growth rate for countries experiencing growth decelerations in 1995-2005 was less than half that in previous decades, contributing to the more positive overall economic performance of the period. Economic declines had both the highest frequency ­ double that of the next highest decade ­ and the greatest impact on countries during the period 1985-1995. Over the entire 30 year period richer countries have had more growth accelerations and poorer countries more growth collapses. This is of course to some 11 extent endogenous; average income per capita will tend to rise in countries with more frequent growth accelerations and fall in countries with more frequent collapses. But this result also holds in each ten year period, where the compounding effects may be assumed to be less important. This may indicate that richer countries are better able to take advantage of propitious circumstances and that poorer countries are less able to avoid bad times. There is one interesting exception. Income per capita for countries experiencing growth accelerations in 1995-2005 is slightly below the average for the region overall, indicating that growth successes have been spreading to poorer countries in the past decade. Table 2 shows the frequency of growth acceleration and deceleration episodes by country category and compares them with the mean. In general, there is no substantial difference in the probabilities of growth acceleration and deceleration episodes for a given country category. But, while geography does not appear to matter, geology and conflict do.10 As might be expected, oil exporters and resource rich countries have more frequent growth accelerations, but somewhat unexpectedly, the same frequency of growth decelerations as the regional average. Conflict is also important in determining good times and bad. Major conflict countries had fewer growth accelerations than the regional average but also fewer decelerations. They also had significantly lower average growth than the regional average. Taken together these results suggest that major conflict countries were trapped in a low level equilibrium. Minor conflict countries have a substantially higher probability of a growth deceleration than the average and are much more likely to experience bad times than good times. Table 2: Frequency of growth acceleration and deceleration by country category Country category Growth acceleration Growth deceleration Frequency (country- Above/below all Frequency (country- Above/below all years) countries' mean years) countries' mean All countries' mean 0.25 - 0.22 - Coastal 0.26 Above 0.22 Equal Landlocked 0.23 Below 0.22 Equal Coastal without resources 0.24 Below 0.23 Above Landlocked without resources 0.22 Below 0.22 Equal Oil exporters 0.29 Above 0.23 Above Non-oil exporters 0.24 Below 0.22 Equal Resource countries 0.30 Above 0.21 Below Non-resource countries 0.23 Below 0.23 Above Major conflict 0.16 Below 0.17 Below Minor conflict 0.19 Below 0.32 Above 10See country assignment in Table A2 in Appendix. 12 Table A3 in the Appendix shows the unconditional probabilities of growth acceleration and deceleration at the country level and the growth rates during these episodes. The gaps between growth rates during accelerations and decelerations at the country level tend to be high, generating the high growth volatility observed in Africa. The high average growth rates observed in many economies during acceleration episodes also show the resilience and capacity of the region's economies to grow when economic and political conditions favor growth. The magnitude of economic contractions during deceleration episodes similarly indicates the severity of the consequences when economic and political conditions are unfavorable. There are 16 countries in our sample that have avoided growth decelerations altogether. Many ­ Botswana, Cape Verde, Equatorial Guinea, Lesotho, Mauritius, Mozambique, Uganda ­ are among the region's top performers in per capita income growth over the three decades, but not all. Burkina Faso, Guinea, Namibia, São Tomé and Príncipe, and Swaziland are not among the region's growth leaders. Avoiding growth collapses is important for long run success at the country level, but is not the only factor contributing to robust long term growth. Seven countries ­ DRC, Eritrea, Gabon, The Gambia, Madagascar, Mauritania and Niger ­ have never had a growth acceleration. Of these only Eritrea shows good long term per capita income growth. Four of the seven had long run declines in per capita income. 5. Do growth accelerations and decelerations matter? We have shown that growth accelerations and decelerations are an important feature of Africa's low and volatile long run growth, but do they matter for economic and social outcomes beyond their direct consequences for the rate of growth? If growth accelerations and decelerations have non-neutral impacts, one would expect that economic, social, and governance indicators will be different during such episodes than during normal times. In this section we investigate the hypothesis of non-neutrality of growth volatility by examining differences in mean values in countries experiencing growth acceleration and deceleration episodes and simple correlations between changes in key economic and social variables and the presence or absence of growth 13 accelerations/decelerations. Table 3 shows sample averages during growth accelerations, decelerations, and "normal" times. ­ defined as the absence of either. Table 4 gives the correlation coefficients between a number of economic, social governance and institutional characteristics and the frequency of acceleration and deceleration episodes. Table 3 reveals an asymmetric relationship between growth accelerations and decelerations and some economic indicators. The major changes in national accounts during growth episodes take place in investments and savings rather than in consumption. Savings and investments are higher during accelerations as compared with normal times, and substantially lower during deceleration episodes. Foreign direct investment during accelerations is six-times the figure for deceleration episodes. Table 4 shows that countries that have high savings and investment have a higher probability of growth acceleration and less probability of deceleration. Consumption is relatively lower during growth accelerations, which is consistent with the higher allocation of resources for investment. But consumption is also lower during decelerations, which is probably due to the fall in purchasing power of households. With regard to the structure of the economy, the share of the agriculture sector is slightly higher in countries experiencing decelerations, while industry's share is slightly larger in countries going through accelerations. Correlations suggest that countries that rely more on agriculture have more spells of growth deceleration, possibly because of higher exposure to insects, draughts and other natural disasters, but also because of swings in agriculture commodity prices. Macroeconomic management appears to be an important factor in both good times and bad times. Decelerations are accompanied by high inflation; one recent example is Zimbabwe. There is a positive correlation between inflation and the frequency of growth decelerations. Public debt is higher during both acceleration and deceleration episodes than during normal times, and government consumption falls during both accelerations and decelerations. Correlations suggest that countries that increased their debt also experienced more growth accelerations and decelerations, which may support the view that prudent debt management is important for reducing growth volatility. 14 15 tset-t * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * on atireleced e ) rwis 45).( (.36 34).( 26).( 62).( 46).( 44).( 40).( .3)1( 66).( 82).( .21)3( .19)1( (.49) 52).( .3)2( .4)5( .81)5( .15)3( 29).( 42).( 00).( .68)4( ,71)2( )4.3 (1 03).( (.75) he Ot 12.7 21.9 12.8 .053 91.9 29.4 25.8 44.7 75.1 30.4 44.5 24.61 07.31 .85- 4.111 61.2 23.2 .732 5.29 16.8 51.0 92. 48.81 .558 .9 52 .183 .56- )29.(64.- (.60) (.64) (.64) (.65) 00).( 00).( .62- 56.- .64- .64- 09. 09. owthrG )1 ) 5). .6) 6). .3) ) ) ) ) ng (.66 43).( )2 )4 )9 )5 )4 )6 )8 )3 )9 )9 )1 )7 )8 9)9 6)9 )6 (.15 (.74 (.89 (.76 (.87 (.81 (211 (2 (5 39).( 38).( 74).( 52).( 52).( .4 .3 (.02 (.01 Duri 7.4 5.91 (.40.29 .77 89. 31. 24. (.653.4 (1.88.5 39).( (.00 26. 32. 183. 114 .95- (.691.1 (1.51.4 (10 (981 771 168. 115 (.315.1 (.618.4 (8.1 (4.2 .93 187. 113. .052(4.14 (.07.72 07.1- )07(.60 -1. .011- .94- .111- .111- 15. .07 t es-tt * * * * * * * * * * * * * * * * * * seg on raeva atirel ce lep ac esiwre )3 32).( 29)(. )7 )9 )9 )3 )7 26) )3 )7 .4) )1 04) .9) .26) 84) )8 )1 )0 .93) 92) )4 )1 )1 22 (.17 (.5 (.4 (.4 (.3 .3 .4 .0 .5 (1.1 (.6 10.3)( (1 32).( (.5 (.2 (.4 74).( 59).( 66).( 67).( 61).( .0 .9 .6 (30 (2.5 .7 .1 (49. (2.9 )14.1( (.00 m sa n ee htworG Oth (.42.01 9.41 11. 1.9 92 30 24 44 70. 29 (.79.04 551 101 5.9- 13 (2.23.5 (3225.7 69.5 95. 16 50 (.03.9 161 93. .994 3.1 .72- )79(.65.- .71- .67- .71- .61- (.01.1 (.00.1 g )7 )1 ) ) ) ) ) ) ) ) tw rin (.7 (.7 66).( 63).( 99).( 91).( 82).( 83).( .29)2( .15)1( .4)1( .053( .3)1( ) (.84 79).( .4)4( .6)1( .80)1( .726( 47).( 80).( 00).( .338( .58)4( )4.1 04).( (.71 (.58 (.65 (.62 (.57 01).( 01).( bes Du 5.21 3.72 13.71 4.12 88.5 28.6 26.9 44.3 75.4 31.6 43.1 114.4 102.5 .75- 13.6 68.3 15.1 16.6 112.0 15.9 51.3 .91 08. 05. 145.8 84.3 (25.25 3.19 .46- )48.(64.- .59- .46- .66- .57- ence er Diff 3: )hgih=6ot l. Table wol veel P)DG oup)rg 1=elacs(t )hgihot %5 theta ant (% 00)1 P)DG P) P) GD agetnavleer wol .5)2+ to significl GD %( P)DG 000=2( ) ua GDP) de (% (% etraegna ) of ).5 ) %( )shrtbievli %( +2 eqt 100 )P )yc )yc GD )$ (% 1,000 x I) 1,000 enmssessAlanoitutits to5. .5)2+ en en noera .sesehtnre GDP) GDP) ded 000= (2 (US ) GN noitpmusn )srae (y teran (-2 to pa GDP) (% %(tnemtsevni owlftenstnemtsevni GDP) GDP) %(tnu denieic %( Indnayc .5)2+ot5.2(-noitpur ansem (% st GDP) (% (% adert %( 2.5-( co lityib GDP) pr oitaryc letiopm cor co Poli sta .5)2+ot5.2(-ytilauqy quref(tcilf quref(tcilf inrorre sgni enmt atht (% st act atpiacr (% ortalf ycnact pe de pe endn erp(ytialtrom5 .5,2+ot52.(-ytilibatnuoccad of rya al wla con st rd avS esvnI ortcesetaiv Pr ctredingieroF (%noiptmusnoC addeulaverultucirgA adeluavyrts da dunI ddedaeulaveicvreS edarT srtopxE pormI chxeevitcefeflaeR ofsmreT enrruC ADO ADO ermusnoC PDG btedcibl Pu cotnemnrevoG exefiL peeD dernU erp(ytialtromtnafnI jor s:te rimP rytnuoC anecioV liticoP 2.5(-ssenevitcefeftnemnrevoG oratlugeR ofeluR roltnoC conron t-te) Mi Ma No (* antS 16 e nc N na 60,01 02,21 27,01 47,21 01,21 20,21 0 221, 20,21 26,01 132,1 751,1 491,1 2121 13,21 2 211, 452 640,1 0561, 122,1 21,21 512 13,21 248 267 451 104 703 703 307 703 305 295 43,21 342,1 ervog no and ratie 05.