On the Effects of InterSector Productivity Growth Differences on Sectorial Inflation in the Baltic States Relative to the Euro Area

The ability of the Baltic States to damp the growth of inflation might partly depend on the scale of the Balassa-Samuelson (BS) effect. In recent years, inflation was lower in the Baltic States as compared to the beginning of the transition period. This raises the question whether the BS impact on inflation is still relevant. Based on an empirical assessment, this paper aims to provide some recent evidence concerning the productivity growth pattern in the tradable and non-tradable sectors in the new Member States vis-a-vis the euro area. Thus, the paper analyses whether the main assumptions in the Ba lassaSamuelson effect of a higher growth of productivity in the tradable sector and a higher wage growth in the non-tradable sector have come true.


Introduction: links between inflation and productivity and why they are important for the Baltic States
The differences in productivity growth in the tradable and non-tradable goods sectors in the new Member States have frequently emerged in the economic debates as a significant reason why inflation in the new Member States might be higher than in the euro area. This effect was broadly addressed as the Balassa-Samuelson (BS) effect. As productivity grows faster we expect the productivity gap between the tradable and non-tradable sectors to widen. This will put up wage pressures in the nontradable sector, leading to increased inflation. This aspect has been widely analysed in the literature as the BS effectl.
Definition: The Balassa-Samuelson effect is based on the fact that productivity in the tradable goods sector tends to rise relative to productivity in the non-tradable sector as part of the process of real income convergence. Higher real wages in the tradable sector will add pressure on wages in the non-tradable sector, as competitive pressures within the labour market occur. Accordingly, more intensive productivity growth in the tradable goods sector will intensify the relative cost of production in the non-tradable sector.
Thus, the relative price of non-tradables will rise. Given that the productivity in the non-tradable goods sector is usually lower, an excess increase in wages compared to the productivity growth would be followed by higher price increases in the non-tradable goods sector and appreciation of the nominal or real exchange rate. Restriction: The validity of the assumption regarding crosssector wage equalisation is not analysed in this note. However, according to the empirical evidence, there is no significant gap in industry and sector wage levels in the new Member States.
The second effect of productivity growth is a change in the price setting mechanism. The price setting mechanism under imperfect competition implies that firms set prices as a markup over unit labour costs 2 • An increase in productivity increases product variety, which may lead to a mark-up reduction due to a competition effect. This is relevant for the Baltic States, which were transforming towards the introduction of more competitive market structures. The final impact will depend on the relative strengths of the two effects. Given data shortages, I focus mainly on the first (BS) effect in my analysis. In the analysis, the proxies are applied: manufacturing for the tradable goods sector and services for the non-tradable sector. It should be noted, however, that some services are becoming increasingly tradable. Sector productivity is measured as a sector value added in constant prices per person employed (labour productivity). Thus, the ability of new Member States to meet the Maastricht inflation criterion for participation in EMU might partly depend on the scale of the BS effect.
Finally, it is important to note the problems related to the productivity measures in the Bal-2 See MacDonald and Ricci (2002). 8 tic States. One of the applied measures of a country's sector productivity performance is labour productivity -gross value added per person employed by a sector. However, labour productivity is influenced by capital deepening, which increases with the quantity of capital per unit of labour input. Measures of labour productivity take into account the impact of changes in employment on output, but they do not take into account the impact of changes in.the capital stock. Another productivity measure, the total factor productivity, captures the efficiency with which both labour and capital inputs are used, and thus it is the most advantageous measure of productivity. However, the total factor productivity is difficult to measure in the Baltic States, as estimates of the effective capital stock are subject to considerable uncertainty, especially in the earlier stages of the transition period. Thus, labour productivity is applied for the analysis below.

The overview of studies on the BS effect in the Baltic States
The calculation of the BS effect in different studies has led to quite different results. Ac-cordi~g to the literature, the estimates of the BS effect in the Baltic States vary from zero to 2 percentage points per annum 3 • Most studies investigating the BS effect in the Baltic States apply a measure of prices of tradable / non-tradable goods relative to the productivity gap in tradable and the non-tradable sectors. The remaining studies focus on the evolution of real effective exchange rates relative to the productivity gap. It should be noted that this approach may lead to an inaccurate measurement of the BS effect in the Bal-tic States relative to the euro area, as the real effective exchange rates include inflation differentials as well as nominal exchange rate changes vis-a-vis countries outside the euro area.
Here, some general difficulties related to the measuring of the BS effect should be mentioned. This applies mainly to the difficulties to isolate the BS effect from the other factors influencing inflation. In this context, especially country-specific factors such as exchange rate policies, trade structure and investment demand can be mentioned. For instance, under a fixed exchange rate regime the BS effect will result in inflation and real exchange rate appreciation. Under a flexible exchange rate regime it will result in some combination of nominal appreciation and inflation. Moreover, some significant administrative price changes in the service sector have occurred in the Baltic States over the recent years. The recent administrative price change was an increase in heating costs due to the imported gas price inc-

Productivity gaps in the Baltic
States: tradable and non-tradable sectors relative to the euro area from 1996 to 2004 According to Chart 2, the labour productivity catch-up, particularly in the tradable sector, was substantial in the Baltic States in the period 1996-1998. Hence the underlying BS assumptions of a productivity gap between tradable and non-tradable sectors came true. Several factors contributed to the higher productivity growth in the tradable sector in this period. During the transition period, liberalisation and movements in relative prices led to the reallocation of capital resources toward the exporting manufacturing sector. Especially in the first part of the transition period the manufacturing sector attracted most of the FDI, which had a positive impact on capital endowment and technological improvement.
Also, the process of labour shedding in the manufacturing sector as well as the re-allocation of labour between the manufacturing and service sectors led to the inter-sector productivity growth differences. The employment share of the tradable sector was rapidly decreasing in the Baltic States in the period 1996 through 1999 (Chart 2). In the same period, these countries experienced large inter-sector productivity growth gaps.
The inter-sector productivity growth gap in the Baltic States stayed broadly unchanged in the period from 2000 to 2004 4 • Compared to other new EU member states, the Baltic States experienced a rapid real GDP growth in this period. Accordingly, the inter-sector productivity growth differentials stayed above the euro area level also in the period from 2000 to 2004 (see Chart 4). According to Chart 5, wage growth in both sectors was broadly equal in the Baltic States and the euro area. This supports the hypothesis that the unit labour costs (the difference between wage growth and the productivity growth) is higher in the service sector. This is also in line with the fact that the service sector inflation exceeded that of the industrial sector in the period from 1998 to mid 2004. Disregarding the inter-sector productivity growth differentials, it is important to remember that the overall demand pressure coming from the wage increases in the service sector spillover also to the industrial sector inflation. The recent increase in the industrial sector inflation is also due to the exceptional factors such as an increase in oil prices.

Conclusions
In the period from 1996 to 2004, the intersector productivity growth differences (tradable versus non-tradable sectors) were bigger in the Baltic States compared to the euro area. Thus, the convergence in terms of inter-sector productivity growth structure was quite minor. At the same time, a higher growth in both sectors' productivity suggests that there has been a strong convergence in productivity growth in the Baltic States relative to the euro area in terms of the productivity level. Accordingly, the findings suggest that the underlying assumptions for the BS effect have come true.
Thus, it can be expected that a higher service sector inflation will contribute stronger to the overall inflation in the Baltic States compared to the euro area. Disregarding the inter-sector productivity growth differentials, it is important to remember that the overall demand pressure coming from the wage increa- ses in the service sector spills over also to the industrial sector inflation. However, it is very difficult to measure the exact impact of the BS effect on inflation. This applies mainly to the difficulties to isolate the BS effect from other factors influencing inflation. Moreover, there is no clear trend that inflation in the non-tradable sector exceeds that in the industrial sector since 1998. However, the non-tradable sector inflation has stayed above that in the tradable sector since 2005 in the Baltic States.
Thus, when analysing the pass-through from productivity growth to inflation in the Baltic States several caveats should be kept in mind: 1) the share of non-tradable goods in consumption is still significantly lower in the Baltic States than in the euro area. The estimations suggest that the share of non-tradables in consumption is just above one third in the Baltic States, which is lower than in the euro area countries. Thus, it limits the magnitude of the pass-through from the productivity growth to inflation in the Baltic States compared to the euro area. Moreover, the BS effect will tend to be smaller in small open economies, such as the Baltic States, as the share of tradable goods in consumption is relatively high; 2) factors other than the inter-sector productivity gaps contributed to the higher inflation in the non-tradable sector in most ofthe Baltic States in 1998 through mid 2004. For instance, an increase in the administrative prices, which cover a significant share in consumer price indices in the Baltic States and are mainly concentrated in the service sector, contributed significantly to the increase in the price of non-tradables. The recent administrative price changes included an increase in heating costs in the Balties due to the imported gas price increases, mainly during the last year. Also, the change in the structure of demand might have been an important factor, as the demand for services tends to increase as real income rises. In this context, it is important to note that administrative prices are typically concentrated in the non-tradable sector. Moreover, price liberalisation has contributed directly to an increase in the price of non-tradables. Nevertheless, the price level, most particularly in the nontradable sector, is still considerably lower in the Baltic States compared to the euro area.

Summary
In the period from 1996 to 2004, inter-sector productivity growth differences (tradable versus non-tradable sectars) were bigger in the Baltic States compared to the euro area. Thus, the convergence in terms of inter-sector productivity growth strueture was quite minor. At the same time, the higher growth in both sectors' productivity suggesls that there has been a strong convergence in productivity growth in the Baltic States relative to the euro area in terms of the productivity level. In line with developments in the euro area, sector wage growth differences were small in the Baltic States. At the same time wage growth in both sectars was much stronger than in the euro area.
Thus, tbe findings suggest that the underlying assumptions for the Balassa-Samuelson etfect are ful-fIlIed. Accordingly, it can be expected that higher service sector inflation will contribute stronger to the overall inflation in the Baltic States compared to the euro area. According to the literature, the estimates of the Balassa-Samuelson etfeet in the Baltic States vary from zero to 2 percentage points per aonum. Disregarding the inter-sector productivity growth differentials, it is important to remember that the overaU demand pressure coming from the wage increases in the service sector spill-over also to the industriai sector inflation.
However, it is very ditficult to measure the exact impact of the Balassa-Samuelson etfeet on inflation. This applies mainly to the diffieulties to isoiate the Balassa-Sarnuelson etfect from other factors influencing inflation. Moreover, there is no elear trend that inflation in non-tradable sectars exceeds that in the industriaI seetor since 1998. However, non-tradable sector inflation stayed above the tradable seetor since 2005 in the Baltic States.
Thus, when analysing the pass-through from productivity growth to inflation in the Baltic States, severai caveats have to be kept in mind. The share of non-tradable goods in consumption is stiU significantly lower in the Baltic States than in the euro area. Thus, it limils the magnitude of the pass-through from the productivity growth to inflation in the Baltic States compared to the euro area. Moreover, the Balassa-Samuelson etfect will tend to be smaIIer in small open economies such as the Baltic States, as the share of tradable goods in consumption is relatively high. Factors other than the inter-sector productivity gaps contributed to the higher inflation in the non-tradable sector in most of the Baltic States from 1998 to mid 2004. For instance, an increase in the administrated prices, which cover a significant share in consumer price indices in the Baltic States and are mainly concentrated in the service sector, contributed significantly to the increase in the price of nontradables. The recent administrative price changes were the increase in heating costs in the Baltics due to the imported gas price increase mainly during the last year. Also, the change in the structure of demand might have been an important factor, as the demand for services tends to increase as real 16 income rises. In this context, it is important to note that administered prices are typically concentrated in the non-tradable sector. Moreover, price liberalisation has contributed directly to the increase in the price of non-tradables. Nevertheless, the price level, most particularly in the non-tradable sector, is still considerably lower in the Baltic States compared to the euro area.