Abstract
This paper estimates the magnitude of the Balassa-Samuelson effect for Greece. We calculate the effect directly, using sectoral national accounts data, which permits estimation of total factor productivity (TFP) growth in the tradeables and nontradeables sectors. Our results suggest that it is difficult to produce one estimate of the BS effect. Any particular estimate is contingent on the definition of the tradeables sector and the assumptions made about labour shares. Moreover, there is also evidence that the effect has been declining through time as Greek standards of living have caught up on those in the rest of the world and as the non-tradeables sector within Greece catches up with the tradeables.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
The BS effect may exist irrespective of the exchange rate regime. For countries with flexible exchange rates, which are catching up, the BS effect will again imply real appreciation without the appreciation involving a loss in competitiveness. For countries that are planning on joining a monetary union or some form of exchange rate system (such as ERM II), some measure of the extent of the BS effect may be helpful in setting the exchange rate at which entry will take place. Many papers deal with the extent of the BS effect in new EU Member States (De Broeck and Slok 2001; Arratibel et al. 2002; Flek et al. 2002; Fischer 2002; Mihaljek and Klau 2003; Crespo-Cuaresma et al. 2005; Egert et al. 2006). See Doyle et al. (2001) for a review of earlier studies.
See, for example, the extensive discussion in Ito et al. (1997) on the particular relevance of the BS effect for fast growth economies. They argue that it is in such countries that the differentials in productivity increases are greater.
The vast majority of the literature, whilst recognising that it is total factor productivity differences between the two sectors that determine the size of the BS effect, uses relative labour productivity in empirical analyses. An exception is Katsimi (2004).
Evidence supporting the fact that the BS effect is insignificant in more highly developed countries comes from the strong association between real exchange rate appreciation and economic development (Ito et al. 1997; Drine and Rault 2003, 2005). In the literature that actually considers the BS effect in developed countries, the evidence about its existence is mixed. One problem preventing a more definitive conclusion is that often panel data techniques are used (and hence the evidence in favour of a BS effect is a general average for all countries in the sample) and/or the actual size of the effect is not calculated. See Faria and Leon-Ledesma (2000), Ortega (2003), Katsimi (2004) and Drine and Rault (2005).
That is, \(\hat P = \gamma \hat P_T + \left( {1 - \gamma } \right)\hat P_N \), where \(\gamma = {{Y_T } \mathord{\left/ {\vphantom {{Y_T } Y}} \right. \kern-\nulldelimiterspace} Y}\) and \(\left( {1 - \gamma } \right) = {{Y_N } \mathord{\left/ {\vphantom {{Y_N } Y}} \right. \kern-\nulldelimiterspace} Y}\).
In contrast, the international BS effect is: \( \widehat{P} - \widehat{P} = {\left( {1 - \gamma } \right)}{\left[ {\widehat{P}_{N} - \widehat{P}^{ * }_{N} } \right]} \), where the “stars” refer to corresponding foreign values of aggregate inflation and sectoral inflation differentials respectively. In the literature, the domestic version is also referred to as the Baumol–Bowen (Baumol and Bowen 1966) effect.
Other mechanisms, however, could also generate such a result including systematic differential productivity growth combined with centralised wage bargaining, which tends to lead to similar increases in wages across sectors irrespective of sectoral productivity developments.
A variation on this theme is provided by Arratibel et al. (2002), who estimate separate equations for domestic inflation, tradeables inflation and non-tradeables inflation augmented by productivity in manufacturing.
A variation on the real exchange rate equations is the paper by Crespo-Cuaresma et al. (2005), which estimates a monetary model of the nominal exchange rate extended to include the price of non-tradeables relative to tradeables. A non-econometric approach is used by Ortega (2003) and ECB (2003), under which inflation differentials are decomposed into the purchasing power parity condition in the traded goods sector along with mark-ups, nominal wages and labour productivity differentials between the nontraded and traded goods sectors.
The size of the BS effect for the more recent period is calculated using the parameters estimated over the whole period along with productivity differentials and inflation differentials for the period 1990–1996. Swagel (1999) defines the tradeables sector as including mining and quarrying and manufacturing. Agriculture is excluded from the sample altogether.
Unfortunately, a lack of data on gross fixed capital formation by sector from 2004 makes updating impossible.
Administered prices are not just a feature of agriculture. Other goods and services are also subject to regulation and there is evidence that prices in these sectors behave differentially from other services (Lunnemann and Matha 2005; Egert et al. 2006). In the euro area, administered prices (for both goods and services and whether they are “fully” or “mainly” administered) account for around 13.8% of the Harmonised Index of Consumer Prices (HICP). Fully administered prices have a weight of 3.4% in the HICP. In Greece, the figures are similar at 12.3 and 4.5%, respectively. This compares with around 15–25% in most transition economies (Egert et al. 2006, Table A1).
Depreciation rates by industry and type of capital good are given by Timmer and O’Mahony (2005) and they show that whilst differences across industries do exist, the main difference is across types of capital good. For example, Timmer and O’Mahony report depreciation rates across all industries of 13.2% for non-ICT equipment and 2.8% for structures. Since it might be expected that the tradeables sector has a higher proportion of machinery and equipment than the nontradeables sector, it is likely that the capital stock of the tradeables sector depreciates more quickly. Unfortunately we do not have data on gross fixed capital formation by type of capital good for each industry to provide evidence that our assumption is borne out. However, experiments with other depreciation rates (including identical rates for both sectors) do not indicate that our results are sensitive to the depreciation rates chosen here.
The base effect in calculating the capital stock arises because we do not actually have an estimate for the capital stock at any point in time. Hence we assume an arbitrary starting value (in our case, the level of gross fixed capital formation in 1948). In each year the capital stock depreciates and gross fixed capital formation is undertaken. As a result, over time, our assumed starting value contributes an ever smaller amount to the calculated capital stock figure at any point in time. Thus, after a certain amount of time, the actual starting value chosen will have effectively no impact on the capital stock figure.
Tradeables wage growth is average blue-collar hourly wage growth in manufacturing and mining and quarrying weighted by the shares of gross value added in each sector. Non-tradeables wage growth is average rates of growth of monthly wages in retail trade, banks and insurance as well as the hourly minimum wage for blue-collar workers. The results are robust to other definitions of the two sectors. We choose to present the results in Fig. 3 because these measures generate the longest period of data.
We take 5 or 10-year moving averages of output growth, labour force growth, capital stock growth and the share of nontradeables in total production and then calculate TFP growth rates and the BS effect itself.
For reasons of space, the results are not presented here. They are available, on request, from the authors.
As a check of the confidence we can have in these new BS estimations using the truncated data, we begin by estimating the BS effects for the definitions of tradeables and nontradeables used for the whole dataset. The reason why our estimates of the BS effect might differ from the truncated dataset compared to the whole dataset has to do with the method of calculating the capital stock. It should be recalled that to minimise the impact of assuming an arbitrary starting value for the capital stock (in this case gross capital formation in 1948), we calculate the capital stock for the whole sample starting from 1948. With data on gross fixed capital formation for the new sectoral definitions only being available from 1988, the base effect remains in 1995. Hence starting points are based on average capital/output ratios for the whole dataset for tradeable and nontradeables (that is, around 1.5 for tradeables and around 3.6 for nontradeables). As a cross-check of this assumption, we calculate the capital stock using this method for the three definitions of tradeables and nontradeables used up until now and compare with capital stock figures calculated from 1948. This confirms that we get sensible results for the capital stock growth figures, which are an important input into calculating TFP in tradeables and nontradeables. The new BS estimates confirm that the three definitions of tradeables generate similar results whether we use the whole dataset or the truncated one. If tradeables are defined to include mining and quarrying and manufacturing, then the average BS effect 1996–2003 using the full data set (and differential labour shares) is 0.22 compared with 0.48 for the truncated data set; including agriculture in tradeables produces BS effects of -0.06 and 0.08, respectively; including transport and communication in tradeables produces BS effects of 0.65 and 0.69, respectively. This suggests that we can have quite a bit of confidence in the following results, which redefine the tradeables sector even more broadly.
We use Swagel’s (1999) results because he also uses total factor productivity and they are, therefore, comparable with our methodology.
Trade weights are calculated using the IMF’s Direction of Trade Statistics. We calculate the bilateral trade weights as the sum of Greece’s exports and imports to/from country i divided by the sum of Greece’s total exports and imports to, and from, the EU. Trade weights are averages over the period 2000–2003.
The lack of data on wage growth prevents a similar test for wages being conducted.
References
Alberola-Ila E, Tyrvainen T (1998) Is there scope for inflation differentials in EMU? Banco de Espana, Working Paper, no. 9823
Arratibel O, Rodriguez-Palenzuela D, Thimann C (2002) Inflation dynamics and dual inflation in accession countries: a “New-Keynesian” perspective. ECB Working Paper, no. 132
Balassa B (1964) The purchasing power parity doctrine: a reappraisal. J Polit Econ 72:584–96
Baumol W, Bowen W (1966) Performing arts: the economic dilemma. New York: 20th Century Fund.
Bragoudakis Z, Moschos D (2000) Relative prices and sectoral labor differentials: a long-run analysis for Greece. Ekonomia 4(2):142–55
Brissimis S, Sideris D, Voumvaki F (1998) Purchasing power parity as a long-run relationship: an empirical investigation of the Greek case. Bank of Greece Economic Bulletin 12:21–54
Brissimis S, Sideris D, Voumvaki F (2005) Testing long-run purchasing power parity under exchange rate targeting. J Int Money Financ 24(6):959–81
Canzoneri M, Cumby R, Diba B, Eudey G (2002) Productivity trends in Europe: implications for real exchange rates, real interest rates and inflation. Rev Int Econ 10(3):497–516
Crespo-Cuaresma J, Fidrmuc J, MacDonald R (2005) The monetary approach to exchange rates in the CEECs. Econ Transit 13(2):395–416
De Broeck M, Slok T (2001) Interpreting real exchange rate movements in transition countries. IMF Working Paper, WP/01/56
Doyle P, Kuijs L, Jiang G (2001) Real convergence to EU income levels: Central Europe from 1990 to the long term. IMF Working Paper, WP/01/146
Drine I, Rault C (2003) Do panel data permit the rescue of the Balassa–Samuelson effect for Latin American countries. Appl Econ 35:351–59
Drine I, Rault C (2005) Can the Balassa–Samuelson effect explain long-run exchange rate movements in OECD countries. Appl Financ Econ 15:519–30
ECB (2003) Inflation differentials in the Euro area: potential causes and policy implications. ECB, Frankfurt
Egert B, Halpern L, MacDonald R (2006) Equilibrium exchange rates in transition economies: taking stock of the issues. J Econ Surv 20:257–324
Faria JR, Leon-Ledesma L (2000) Testing the Balassa–Samuelson effect: implications for growth and PPP. University of Kent, Discussion Papers in Economics, 00/08
Fischer C (2002) Real currency appreciation in accession countries: Balassa–Samuelson and investment demand. Deutsche Bundesbank, Discussion Paper, 19/02
Flek V, Markova L, Podpiera J (2002) Sectoral productivity and real exchange rate appreciation: much ado about nothing? Czech National Bank, Working Paper Series, 4
Froot K, Rogoff K (1995) Perspectives on PPP and long-run real exchange rates. In: Grossman G, Rogoff K (eds) Ch. 32 in Handbook of international economics. vol. III. Elsevier, Amsterdam
Ito T, Isard P, Symansky S (1997) Economic growth and real exchange rate: an overview of the Balassa–Samuelson hypothesis in Asia. NBER Working Paper, no. 5979
Katsimi M (2004) Inflation divergence in the Euro Area: the Balassa–Samuelson effect. Appl Econ Lett 11:329–332
Lommatzsch K, Tober S (2006) Euro-Area inflation: does the Balassa–Samuelson effect matter? International Economics and Economic Policy 3(2):105–136
Lunnemann P, Matha TY (2005) Regulated and service prices and inflation persistence. European Central Bank, Working Paper no. 466
Malley J (2006) The theory and estimation of the Balassa–Samuelson effect: some background notes. (Mimeo)
Mihaljek D, Klau M (2003) The Balassa–Samuelson effect in Central Europe: a disaggregated analysis. BIS Working Paper, no. 143
Obstfeld M, Rogoff K (1996) Foundations of international macroeconomics. MIT Press, Cambridge
Ortega E (2003) Persistent inflation differentials in Europe. Banco de Espana Working Paper no. 0305
Samuelson P (1964) Theoretical notes on trade problems. Rev Econ Stat 46:145–54
Sinn H-W, Reutter M (2001) The minimum inflation rate for Euroland. NBER Working Paper No. 8085
Swagel P (1999) The contribution of the Balassa–Samuelson effect to inflation: cross-country evidence. IMF Staff Country Report, no.99/138
Timmer M, O’Mahony M (2005) Measuring Capital Input. EPKE Technical Note No. 1, National Institute of Economic and Social Research, available at http://www.niesr.ac.uk/research/epke/working_papers.html
Acknowledgments
We would like to thank a number of people at the Bank of Greece who have helped in preparing this paper. Presentations to the Monetary Policy Council yielded useful insights and comments and we would particularly like to thank George Demopoulos. Additionally, thanks are due to George Tavlas who commented on the paper at length as well as several colleagues who took time to discuss data issues with us, including Nick Zonzilos, Isaak Sampethai, and, especially, Daphne Nikolitsas. We would also like to thank Ioannis Papadogeorgis for his help in obtaining data from the OECD. Finally, an anonymous referee helped to sharpen the focus of the paper and made helpful suggestions for improvement. The opinions and analysis presented in the paper are those of the authors and do not necessarily coincide with those of either the Bank of Greece or the Eurosystem.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
1.1 Data appendix
The data was compiled using sectoral national accounts data for Greece from 1948 to 2003. The major problem associated with the data compilation was the many changes in the system of recording national account aggregates, especially at the sectoral level. Whilst, the National Statistical Service of Greece has produced national account aggregates for the whole economy using ESA95 going back to the 1960s, the sectoral information provided is strictly limited. Aside from the fact that new systems of national accounting entail different ways of counting the various aggregates, there is also the problem that sectoral definitions change. In this appendix we provide details on how the database was constructed.
Table 7 defines the two sectors, tradeables and non-tradeables in terms of the classification scheme used in each of the national accounts systems.
Output is measured by Gross Value Added (GVA). The database contains both current and constant (1995) prices GVA. We started with the figures for 1995–2003 (based on ESA95) to calculate the broad sectors of agriculture, industry (including construction), mining and quarrying, manufacturing and services. Figures are then backdated using the EU’s AMECO database for 1960–1994 (by using rates of change). AMECO does not contain information on mining and quarrying and hence figures are backdated to 1988 using rates of change of GVA in mining and quarrying from National Accounts (ESA79) and to 1960 using rates of change of GDP in mining and quarrying from the revised National Accounts 1960–1997 (ESA79)
Gross Fixed Capital formation data (both current and constant 1995 prices) on a sectoral basis is still only available from 1995 to 1998 on a sectoral level from the published national accounts (ESA95). Data was therefore taken from the OECD’s STAN Sectoral Database for 1995 to 2003. We then backdated using rates of change to 1989 from National Accounts data (ESA79) and to 1948 with National Accounts data (ESA70). Data was collected from 1948 in order to help eliminate base year effects when calculating the capital stock using the perpetual inventory method where depreciation is assumed to be 10% per annum in the traded goods sector and 4% in the nontraded sector.
For employment data, we again begin with the ESA95 data as this is more complete, using as it does not only the labour force survey, but also other sources of information (e.g. better data for agriculture than was used in the past). Using this data we calculate the following broad groups: agriculture, industry (plus construction), manufacturing, mining and quarrying and services. We then backdate using growth rates for theses broad aggregates using OECD data on rates of change of civilian employment. For mining and quarrying (where OECD data is unavailable), we also use census data for the earlier period of the 1960s and 1970s and Labour Force Survey data for the 1980s (Labour Force Survey data for the period before 1981 covered the Athens area only).
The tradeables sector is defined in two ways in the paper. First, we include manufacturing along with mining and quarrying, leaving agriculture, industry (minus manufacturing and mining and quarrying) and services as non-tradeables. Second, we test the sensitivity of these results to redefining tradeables to include agriculture.
Finally, for prices we use sectoral CPI rates. Pre-1990 there were nine categories which were subsequently extended to 12 categories. The categories were combined (as detailed in Table 8) to generate measures of traded and nontraded sectors as defined in the main text using the appropriate weights. When we include agriculture as a tradeable, food and non-alcoholic beverage and alcohol and tobacco are included in tradeables.
Rights and permissions
About this article
Cite this article
Gibson, H.D., Malley, J. The Contribution of Sectoral Productivity Differentials to Inflation in Greece. Open Econ Rev 19, 629–650 (2008). https://doi.org/10.1007/s11079-007-9071-3
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11079-007-9071-3