Finland: Selected Issues and Statistical Appendix

This Selected Issues paper and Statistical Appendix examines inflation and wage dynamics in Finland. The paper discusses the data set used (quarterly data covering from 1975 to 1995) and the statistical properties of the relevant time series. It presents the model and the empirical estimates, and provides an outlook for consumer price index and nominal wage inflation for 1996:Q1–2001:Q4. The paper examines the determinants of the equilibrium real exchange rate, and also analyzes the Finnish banking system and the credit crunch hypothesis.

Abstract

This Selected Issues paper and Statistical Appendix examines inflation and wage dynamics in Finland. The paper discusses the data set used (quarterly data covering from 1975 to 1995) and the statistical properties of the relevant time series. It presents the model and the empirical estimates, and provides an outlook for consumer price index and nominal wage inflation for 1996:Q1–2001:Q4. The paper examines the determinants of the equilibrium real exchange rate, and also analyzes the Finnish banking system and the credit crunch hypothesis.

I. Inflation and Wage Dynamics in Finland: A Cointegration Approach 1/

1. Introduction and summary

Following the abandonment of the markka’s unilateral link to the ECU in September 1992, the Bank of Finland (BoF) committed itself in February 1993 to stabilize underlying inflation at around 2 percent. 2/ This commitment was tested in late 1994 and early 1995, when monetary policy was tightened significantly amidst increased inflation expectations. These expectations were due to the strengthening of aggregate demand, and to the acceleration of wages in the context of the more decentralized wage negotiation mode adopted in the fall of 1994. Inflationary pressures in 1995 weakened (annual CPI inflation was 1 percent), owing to the slowdown of growth, the appreciation of the markka, and the decline in food prices related to the accession in the EU (Chart 1). Starting in October 1995, the BoF has eased monetary policy significantly, citing weak inflationary pressures and a positive wage outlook; the latter reflected the moderate two-year wage agreement reached that month. 3/ The October 1995 agreement marked a return to a centralized wage bargaining framework. 4/

CHART 1
CHART 1

FINLAND: INFLATION, 1975–95 1/

(In percent)

Citation: IMF Staff Country Reports 1996, 095; 10.5089/9781451813128.002.A001

Source: Bank of Finland.1/ Annual growth rates.2/ CPI excluding indirect taxes, subsidies, and housing-related capital costs.3/ Excluding nonwage labor costs.

The wage moderation of the October agreement bodes well for the inflation outlook. However, there is uncertainty regarding the size of the wage drift, as well as the effect on prices of the recent weakening of the markka and the fading away of the effects of EU accession. Against this background, this paper presents econometric projections on consumer price and nominal wage inflation in Finland over 1996–2001. A simple model of inflation and wage dynamics is constructed based on Johansen’s cointegration framework. It is shown that a VAR system consisting of a long-term relationship for CPI inflation and another for real wages and their respective error-correction mechanisms offers a parsimonious description of wage-price interaction.

To preview the results, the main influences on price and wage inflation are as follows. In the long term, CPI inflation is related to nominal wage growth and imported inflation; in the short term, there is also positive feedback from changes in indirect taxes. In the long run, the real wage depends on unemployment, labor productivity, and indirect taxes, while in the short run there is a significant contribution of changes in unemployment and indirect taxes in eliminating deviations from the real wage equilibrium. Changes in unemployment and indirect taxes exert a stronger influence on real wage adjustment than productivity growth. Finally, the impact of cyclical variables on wage and price behavior is found to be insignificant, with the exception of the unemployment rate.

A relevant implication of the model is that an increase in indirect taxes raises real wages net of taxes in the short term, whereas in the long term a negative correlation prevails. This evidence confirms earlier results by Tyrväinen (1995c) in favor of “real wage resistance”; it suggests that high unemployment in Finland may be at least partly explained by the increase in the tax wedge in recent years.

The VAR system is used to project CPI and nominal wage inflation for 1996–2001. The model implies that—under reasonable assumptions on the behavior of the unemployment rate, labor productivity, and indirect taxes—the BoF’s 2 percent inflation target can be maintained as long as the markka remains broadly stable on the levels reached in the first quarter of 1996. Specifically, CPI inflation is projected to peak in early 1997 at about 2 percent under the assumption of no further nominal effective depreciation after the first quarter of 1996. Regarding nominal wage inflation, the outlook is for a moderate increase over the period.

The remainder of the paper is divided into three parts. Section 2 discusses the data set used (quarterly data covering the period from 1975 to 1995) and the statistical properties of the relevant time series. Section 3 presents the model and the empirical estimates. The final section provides an outlook for CPI and nominal wage inflation for the period 1996:Q1–2001:Q4.

2. Data sources and statistical properties

a. Data sources and definitions

The data set employed consists of seasonally adjusted quarterly time series for the past two decades (1975:Q1–1995:Q4). 1/ The key endogenous variables are the consumer price index (P) and the index of nominal wages (W). The latter covers both the public and private sectors and excludes nonwage labor costs, such as social security contributions. 2/ The other variables employed in the final form of the VAR system are the unemployment rate (U), 3/ labor productivity (LPR) measured as the real GDP-to-employment ratio, an indirect tax index including subsidies and consumption taxes (T), and the import price index (PH). Import prices are measured in Finnish markka and are thus affected by both exchange rate and world price changes. Some of these variables enter the model in terms of differences: for example, DP and DW denote the four-quarter difference of P and W. Fourth quarter differences are used to remove seasonality and reduce the volatility of differenced variables. Simplifying the lag structure in this way keeps the number of estimated parameters manageable, albeit at the cost of reducing the short-term interaction among the variables.

b. Order of integration

Before carrying out the cointegration analysis, the relevant variables need to be tested for the order of integration. Augmented Dickey-Fuller (ADF) unit root tests were performed on the logarithms of the time series to determine their stationarity properties (Table 1). As is often the case with nominal variables, the consumer price index and the nominal wage index are each integrated of order two—so that the time series have to be differenced twice to obtain stationarity—while their four-quarter growth rates are integrated of order one. Real wages are nonstationary and integrated of order one. Labor productivity and the indirect tax index are integrated of order one, while their four-quarter growth rates are stationary. The import price index appears to be integrated of order one at 5 percent probability. However, as the power of the test is low and the critical values of the test statistics are typically sensitive to the unknown true dynamic structure of the data generating process, we will treat import prices as being integrated of order two, in order to allow for cointegration between traded goods prices and the exchange rate and domestic prices. 4/ Finally, the unemployment rate is integrated of order zero at 5 percent but integrated of order one at 1 percent probability. This evidence of nonstationarity likely reflects the structural break accompanying the sharp rise in unemployment following the onset of the recession in 1990; it suggests that cyclical output measures are not particularly reliable indicators of macroeconomic policy changes in Finland (see below).

Table 1.

Finland: Results of Integration Tests

article image
Source: Staff calculations.

All series except the unemployment rate are in logarithms.

MacKinnon (1990) 95 percent critical values for rejection of hypothesis of a unit root.

Four-quarter growth rate.

3. Empirical estimates

a. Modeling strategy

When nominal macroeconomic time series are nonstationary 1/, standard estimation methods are inappropriate. Engle and Granger’s (1987) two-step method suggests that there may be a linear combination of the individual nonstationary series which is stationary; this is referred to as a cointegrating vector and is usually interpreted as a long-term equilibrium relationship between the variables. Deviations from the long-term equilibrium are dealt with in the context of an error-correction mechanism (ECM).

This paper adopts the maximum likelihood procedure of Johansen (1988, 1991) and Johansen and Juselius (1990) for constructing a VAR (Vector Autoregression) system permitting the simultaneous consideration of several cointegrating vectors. Johansen’s framework distinguishes between long-term effects and short-term dynamic responses of variables, while allowing all relevant parameters to be estimated simultaneously. Ideally, the data support a unique long-term relationship, deviations from which are gradually eliminated via the error-correction mechanism.

Mühleisen (1995) applied elements of Johansen’s and Engle and Granger’s methodologies in the context of a wage-price model for Finland; Johansen’s maximum likelihood test was first carried out to obtain the long term relationships and the Engle-Granger approach was then used to analyze short-run inflation adjustment. Instead, this paper focusses on Johansen’s methodology and constructs a VAR system allowing the estimation of long-term and short-term relationships for CPI inflation and real wages. In this way, it is possible to separate equilibrium behavior from short-term dynamic responses and avoid the possible single equation bias that would likely disturb the two-stage results.

The simplest cointegration framework for modeling inflation and wage dynamics would posit a single equilibrium relationship for nominal wage inflation as a function of consumer price inflation and the growth rate of labor productivity. An error-correction mechanism would be jointly estimated to handle the system’s dynamic response to deviations from equilibrium.

In implementing this approach two problems arise. First, although a unique cointegrating vector relating the three variables can be identified, the sign of the cointegrating residual in the error-correction mechanism for price inflation fails to have the appropriate (positive) sign. It follows that a deviation from equilibrium is magnified rather than eliminated, leading to inflation dynamics which eventually explode. Second, the estimate of the coefficient of the error-correction term in the EGM equation for the growth rate of labor productivity is significant, suggesting that productivity growth is entirely endogenous in this system, and does not depend on either technical progress or innovation.

These problems appear both in the basic version of the VAR system, in which only the variables in the cointegrating vector enter into the ECM, as well as in several extended versions in which potential explanatory variables enter exogenously into the ECM. Exogenous variables that were used include woodpulp prices—considered a good leading indicator of CPI inflation in Finland (see Mühleisen (1995))—the nominal exchange rate, employers’ social security contributions as an proxy of firms’ nonwage labor costs, expected (imported) inflation, proxied by the first difference of the deviation of the import price index from domestic manufacturing prices, the growth of the money supply (both Ml and M2) scaled by real GDP growth, capacity utilization in manufacturing, cyclical and total unemployment, and the output gap. 1/

On account of these difficulties, a somewhat more complex procedure was followed. Specifically, a VAR system was estimated consisting of one cointegration vector for the consumer price index and another for real wages. and their respective error-correction mechanism. In this way, the VAR system has two distinct attractors, 2/ and the dynamic responses of the cointegrated variables to deviations from equilibrium turn out to be well defined.

In this specification, the real wage is negatively related to the unemployment rate and positively related to labor productivity. 1/ Short-term real wage adjustment is then described by an error-correction mechanism. 2/ As a statistical matter, imposing homogeneity between wages and prices a priori is a desirable restriction given the relatively small size of the data set, as it reduces the dimension of the VAR system. Moreover, Tyrväinen (1995c) in a different context indicated that imposing this restriction improves the overall fit of a wage equation.

As to the cointegration vector for consumer prices, CPI inflation is directly related to the growth rates of nominal wages and the import price index. The long-term relationship may thus be interpreted as a mark-up equation, an approach also adopted by Ford and Krueger (1995) and Mühleisen (1995) for Italy and Finland, respectively. An attempt was also made to estimate a long-term relationship between the CPI and the PPI. However, using Johansen’s maximum likelihood test it was concluded that the consumer price and producer price indices are not cointegrated over the last two decades. This is not too surprising: the large size of Finland’s export sector suggests that fluctuations in the producer price index may not be a good indicator of CPI changes.

b. Cointegration and error correction

The previous discussion is consistent with the statistical properties of the time series. The levels of nominal variables are nonstationary and integrated of order two, with the exception of the indirect tax index, while the levels of real variables are integrated of order one, with the exception of the unemployment rate. Therefore, the variables in the cointegration vector for the consumer price index should enter in first differences, while their dynamic response should be in second differences. Correspondingly, the variables entering the long-term relationship for real wages should be in levels, while the equation for the error-correction mechanism should be in growth rates.

Johansen’s maximum likelihood procedure then yields the following cointegration relationships for CPI inflation and the real wage: 1/

ΔP=0.60ΔW+0.36ΔPM+CIPWP=3.050.02U+0.98LPR0.30T+CIW,

where CIP and CIW are the stationary cointegrating residuals. Each cointegration vector is unique at the 5 percent significance level. As to the real wage equation, the estimated coefficients have the appropriate signs: the level of productivity and the unemployment rate are, respectively, positively and negatively correlated with the real wage. The negative coefficient on indirect taxes suggests that in the long run a higher level of indirect taxes yields a lower real wage. More generally, an increase in the tax wedge is eventually born by labor, as also documented by several cross-country studies. 2/

The error-correction mechanism in the VAR system involves the joint estimation of the short-term dynamic responses of the variables to deviations from the equilibrium. The ECM for the consumer price index is specified in second differences, while that for real wages is in first differences; the latter incorporates a Phillips curve in the form of changes in the unemployment rate. Several cyclical variables potentially influencing the adjustment processes for inflation and real wages—including the unemployment rate, output gap measures, and short-term interest rates—were included but were not found significant. An exception was the indirect tax index, which enters significantly the ECM for GP1 inflation as an exogenous variable.

For each cointegration vector under consideration, the coefficient of the cointegrating residual is significant only in the adjustment equations for CPI inflation and real wages. Deleting the coefficients of insignificant variables in these two VARs yields the following adjustment equations (t-statistics are reported in parentheses):

Inflation adjustment:
Δ2=0.40Δ2P(1)+0.07Δ2T(1)+0.03Δ2PMPI(1)0.23CIP(1)(4.05)(2.27)(1.87)(1.93)N=84(75:1,95:4),R2=0.32,SEE=0.006,SchwartzInfomationCriterion=10.05
Real wage adjusment:
Δ(WP)=0.07Δ(WP)(1)0.01ΔU(1)+0.12ΔT(1)0.05CIW(1)(2.91)(2.43)(2.44)(2.06)N=84(75:1,95:4),R2=0.39,SEE=0.01,SchwartzInformationCriterion=8.53

Constant terms were included in the regressions but were not found to be significant. According to the first equation, short-term fluctuations in nominal wage inflation do not have a significant influence on CPI inflation adjustment, whereas the first lag of CPI inflation does. Fluctuations in real wages are also affected by their first lag, albeit with a small negative sign. Changes in unemployment and indirect taxes have opposing influences: an increase in unemployment lowers real wages, whereas higher indirect taxes contribute to higher real wages in the short term.

The positive impact of higher indirect taxes on real wages is evidence of “real wage resistance.” namely a situation where firms’ real labor costs rise as a result of exogenous changes in the “tax wedge” which reduce workers’ real take-home pay. 1/ Tyrväinen’s (1995c) study of wages, taxation, and employment in Finland has shown that the impact on real labor costs is independent of whether tax revenue is collected in the form of income taxes, indirect taxes, or social security contributions. In the context of this model, therefore, higher indirect taxes lead to a temporary increase in real labor costs and yield higher unemployment in the short term. Real wage resistance is also consistent with the fact that the correction of deviations from the long-term equilibrium is faster for CPI inflation than it is for real wages: the coefficient of the cointegrating residual for real wages (0.05) implies that three years are required for 50 percent of any deviation from the equilibrium real wage to be eliminated. 2/

c. Model multipliers

In order to investigate the properties of the estimated model, four simulations of the four-equation VAR system were carried out, each subjecting the model to a unit shock to an exogenous variable. 1/ The following shocks were considered: a 1 percentage point permanent decline in the unemployment rate; a 1 percentage point permanent increase in indirect taxes; a 1 percentage point permanent increase in import prices; and a 1 percentage point increase in productivity growth for 1996 (leading to a permanent rise in the level of productivity).

The results of the unemployment and indirect tax shocks are shown in Chart 2. Lowering unemployment by 1 percentage point leads to about 1 percentage point higher real wage growth in the first 4 quarters compared to the control solution. Subsequently, nominal wage growth converges to CPI inflation, so that real wage growth remains broadly in line with the control solution. Thus real wage growth eventually returns to its control value; the cumulative increase in the real wage level amounts to slightly over 1 percentage point.

CHART 2
CHART 2

FINLAND: SIMULATION RESULTS, 1996–2001

(Percent difference from control values)

Citation: IMF Staff Country Reports 1996, 095; 10.5089/9781451813128.002.A001

Source: Staff calculations.1/ Annual growth rates.

A permanent increase in indirect taxes by 1 percentage point initially raises real wage growth above the control values, indicating an overshooting of wages with respect to prices. However, after the first four quarters the error-correction term delivers a sharp decline in excess wage inflation, resulting in marginally lower real wage growth compared to the control solution over the five-year projection period. Eventually, the permanent unit shock to indirect taxes by 1 percentage point leads to a cumulative decrease in the real wage level of less than one half percentage point.

The results of shocks in productivity growth and import prices are shown on Chart 3. As the difference of real wage growth from the control solution fades to zero, a one-time increase in the productivity growth rate leads to a cumulative increase in the real wage level by 0.7 percentage points. However, the speed of adjustment is quite slow: complete convergence occurs beyond the five-year simulation period. In the short term, nominal wage and CPI inflation increase marginally compared to the control values, reflecting the fact that productivity growth is not significant in the ECM for real wages.

CHART 3
CHART 3

FINLAND: SIMULATION RESULTS, 1996–2001

(Percent difference from control values)

Citation: IMF Staff Country Reports 1996, 095; 10.5089/9781451813128.002.A001

Source: Staff calculations.1/ Annual growth rates.

A permanent increase in the import price index by 1 percentage point raises both CPI and wage inflation by under one half percentage point against the control values over the first four quarters. Convergence to the control values roughly occurs by the end of the five-year simulation period. However, this multiplier effect should be interpreted with caution, as it suggests that an import price increase is reflected in a decrease in firms’ profits by the same amount. The dynamic response of prices and wages is identical because import prices enter only in the cointegrating vector for CPI inflation, and homogeneity between prices and wages is imposed both in the cointegrating vector and in the error-correction mechanism. 1/

4. Outlook for CPI and nominal wage inflation: 1996–2001

In this section, the estimated model is used to project CPI and wage inflation over the period from 1996:Q1 to 2001:Q4 under different assumptions on exogenous variables.

a. Baseline scenario

Under the baseline scenario, it is assumed that the unemployment rate will decline by one percentage point per year, from 17 percent at end-1995 to 11 percent in 2001. This path falls short of the authorities’ target of 9 percent unemployment by 1999, reflecting the limited progress achieved in 1995–96 in lowering unemployment. The index of indirect taxes is assumed to remain at its end-1995 level. Labor productivity is projected to increase by 1.6 percent in 1996—down from 2.5 percent in 1995 and an average 3.4 percent for 1984–94—reflecting the slowdown in output growth; subsequently it is assumed to increase by 2.7 percent and 3.1 percent per year in 1997 and 1998, respectively, as growth picks up (Table 2).

Table 2.

Finland: Baseline Scenario: Exogenous Variable Projections, 1996–2001 1/

(In percent)

article image
Source: Staff projections.

Indirect taxes are assumed constant at their end-1995 level.

Annual growth rates.

The import price index (in domestic currency) is affected by developments in the exchange rate and traded goods prices. The baseline scenario assumes that the nominal effective exchange rate of the markka will stabilize at its end-March 1996 level (reflecting a nominal effective depreciation of 4.2 percent in the first quarter), with no subsequent change. Traded goods prices are assumed to change according to the projections of export deflators for goods and services for partner countries (measured in US dollars) based on the May 1996 issue of the World Economic Outlook.

The model’s projections for annual consumer price and nominal wage inflation are shown in Chart 4. CPI inflation increases from its end-1995 low to just under 2 percent in 1997:Q1, reflecting the weakening of the exchange rate in 1996:Q1 and the decline in the unemployment rate in 1996. Subsequently, inflation drops to 1.5 percent by 1998:Q2 and then rises smoothly to edge above 2 percent by the end of 2001. The trend increase from 1998 onwards is accounted for by the steady decline in unemployment and the roughly constant rate of nominal effective depreciation.

CHART 4
CHART 4

FINLAND: BASELINE FORECASTS, 1996–2001

(In percent)

Citation: IMF Staff Country Reports 1996, 095; 10.5089/9781451813128.002.A001

Source: Staff calculations.

Annual nominal wage inflation drops from its end-1995 high of 6.2 percent to 4.8 percent in 1996:Q1 and 2–2.5 percent in 1997. Subsequently, annual wage growth rises steadily to about 3 percent in 2001, yielding slightly under 1 percentage point of real wage growth per annum. The sharp decline in 1996 is explained by the process of equilibrium adjustment of nominal wages (see Chart 1) rather than by the trend projections for the exogenous variables. Specifically, the sharp decline in wage inflation upon the onset of the recession in 1990 and the subsequent increase in late 1994 and 1995—due partly to the recovery and partly to the generous wage agreement in 1994—could be viewed as a deviation from the negative trend observed for most of the 1980s.

Wage drift—wage increases obtained at the industry level over and above the negotiated rate—is obtained as the difference between the wage inflation projections and the annualized wage growth specified in the two-year wage agreement of October 1995. In Finland, wage drift has exceeded 2 percent each year from 1970 until 1991, 1/ after which it has remained below 2 percent (Tyrväinen [1995c]). The implications regarding wage drift in 1996 and 1997 are as follows:

Table 3.

Wage Drift Projections, 1996–97

(In percent)

article image
Sources: Ministry of Finance; and staff calculations.1/ Annual growth rate of nominal wages under the two-year wage agreement.2/ Annual growth rate under the baseline projections.

b. Alternative scenario: further depreciation in 1996

This scenario considers the effect on inflation of a further nominal effective depreciation of 5 percent between the second and fourth quarter of 1996. The remaining exogenous variables are unchanged from their baseline projections. The model’s inflation projections are shown in Chart 5.

CHART 5
CHART 5

FINLAND: ALTERNATIVE FORECASTS, 1996–2001 1/

(In percent)

Citation: IMF Staff Country Reports 1996, 095; 10.5089/9781451813128.002.A001

Source: Staff calculations.1/ A further nominal effective depreciation of 5 percent over the baseline in 1996.

On account of the larger depreciation in 1996, annual CPI inflation is considerably higher than under the baseline scenario, peaking at 3.2 percent in the second quarter of 1997. From 1998 onward, CPI inflation stabilizes at just under 2 percent and eventually edges above it.

Nominal wage inflation declines more slowly than under the baseline scenario, reflecting the higher CPI inflation. However, this result should be treated with caution because—as was discussed in the case of the multiplier for import prices in Section 3—import prices do not directly influence real wages, and therefore prices and wages have to move together in the short term.

References

  • Åkerholm, J., and A. Brunila (1994), Inflation Targeting: The Finnish Experience. Paper presented at the CEPR Workshop “Inflation Targets” in Milan, Italy.

    • Search Google Scholar
    • Export Citation
  • Alogoskoufis, G., et al. (1995), “Unemployment: Choices for Europe,” Monitoring European Integration 5, CEPR.

  • Andersen, K. And H.-L. Männistö (1995), “Output Gaps and the Government Budget Balance: The Case of Finland,Bank of Finland Discussion paper 27/95.

    • Search Google Scholar
    • Export Citation
  • Banerjee, A., J. Dolado, J. Galbraith, and D. Hendry, “Co-Integration, Error-Correction, and the Econometric Analysis of Non-Stationary Data,” Advanced Texts in Econometrics. Oxford University Press 1993.

    • Search Google Scholar
    • Export Citation
  • Bank of Finland (1990), The BOF4 Quarterly Model of the Finnish Economy, Bank of Finland, D:73.

  • Brunila, A. (1996), “Estimates of Cyclically Adjusted Budget Deficits: Are They Reliable?,” Bank of Finland Bulletin, 3/96.

  • Engle, R. and C. Granger (1987), “Co-integration and Error Correction: Representation, Estimation, and Testing,Econometrica 55, No.2.

    • Search Google Scholar
    • Export Citation
  • Engle, R., (1991), “Introduction,” in Engle, R. and C. Granger (eds.) 1991, Long-Run Economic Relationships. Oxford University Press.

    • Search Google Scholar
    • Export Citation
  • Ford, R. and T. Krueger (1994), Exchange Rate Movements and Inflation Performance: The case of Italy, International Monetary Fund, Working Paper 95/41.

    • Search Google Scholar
    • Export Citation
  • Johansen, S. (1988), “Statistical Analysis of Cointegration Vectors,” Journal of Economics Dynamics and Control 12, 231254.

  • Johansen, S., (1991), “Estimation and Hypothesis Testing of Cointegration Vectors in Gaussian Vector Autoregressive Models,Econometrica 59, No.6.

    • Search Google Scholar
    • Export Citation
  • MacKinnon, J. (1990), “Critical Values of Cointegrating Tests,University of California-San Diego Discussion Paper, No 90–4.

  • Mühleisen, M. (1995), Monetary Policy and Inflation Indicators for Finland. International Monetary Fund, Working Paper 95/115.

  • Symons, J. and D. Robertson (1990), “Employment Versus Employee Taxation: The Impact On Employment,” in OECD Employment Outlook.

  • Tyrväinen, T. (1995a). “Wage Setting, Taxes and Demand for Labour: Multivariate Analysis of Cointegrating Relations,” Empirical Economics, 20, 271297.

    • Search Google Scholar
    • Export Citation
  • Tyrväinen, T., (1995b), “Real Wage Resistance and Unemployment: Multivariate Analysis of Co integrating Relations in 10 OECD Countries,” The OECD Jobs Study Working Paper Series 5, OECD.

    • Search Google Scholar
    • Export Citation
  • Tyrväinen, T., (1995c), “Wage Determination, Unions and Employment: Evidence from Finland,” Bank of Finland Studies E:3.

1/

Prepared by Demosthenes N. Tambakis.

2/

Underlying inflation is the annual growth rate of the CPI excluding indirect taxes and subsidies and capital costs of owner-occupied housing (Åkerholm and Brunila (1994)).

3/

The tender rate, which is the BoF’s key monetary policy instrument, has been lowered seven times by a total of 250 points over the last nine months.

4/

See Appendix II of “Finland—Recent Economic Developments” (August 21, 1995) and Tyrväinen (1995c) for a discussion of the wage bargaining process in Finland. The two-year wage agreement reached in October 1995 involves a nominal wage increase of 1.8 percentage points over the 12-month period from November 1, 1995, and an increase of 1.3 percentage points over the 12-month period from November 1, 1996. The agreement includes a clause related to wage differentiation, whereby annual wage increases may be reviewed in August 1996 in light of developments in wage drift. In addition, the following escalator clause is specified: if CPI inflation in the 11-month period from August 1995 to July 1996 exceeds 3.1 percent—an annualized rate of 3.4 percent—nominal earnings in August 1996 will rise by the full amount of the difference of the realized CPI inflation from 2.6 percent. Given recent weak inflationary pressures, it is unlikely that the clause will be triggered.

1/

These data were provided by the BoF, which uses these time series in its macroeconometric model (see Bank of Finland (1990) for an earlier version of the model).

2/

Nonwage labor costs are included separately; see below.

3/

As measured by Statistics Finland.

4/

The same problem was encountered by Ford and Krueger (1995) in the case of Italy.

1/

The variance of a nonstationary series changes over time. In general, a data generating process is strictly stationary if all its moments are constant over time.

1/

The output gap is computed as the ratio of actual to potential output; the latter was obtained using the Hodrick-Prescott filter.

2/

An attractor is the space defined by the cointegrating relationship, to which the system will converge. This attractor may correspond to certain macroeconomic equilibria (see Engle and Granger (1991)).

1/

It might seem that a measure of cyclical unemployment (such as the output gap) would be more appropriate than actual unemployment in explaining movements in real wages. Indeed, the concept of cyclical unemployment has been shown to yield a long-run Phillips curve for the United States and several European countries (Alogoskoufis et al. (1995)). In Finland, however, the severity of the recession of 1990–93 and the associated sharp rise in unemployment make it difficult to estimate potential output and the output gap (Andersen and Männistö (1995), Brunila (1996)). Actual unemployment thus seems to be the more appropriate measure.

2/

This approach is also taken in Ford and Krueger (1995) and Pujol and Griffiths (1996).

1/

An intercept is included in the cointegration test for the real wage. The t-statistics are not reported as the distributions of the estimated coefficients are nonstandard.

1/

The tax wedge is defined as the difference between the relevant real wage for the employer (real labor cost) and the relevant real wage for the trade union (real take-home pay). It consists of income taxes, indirect taxes, social security contributions, and the difference between the consumer and producer price indices.

2/

The considerable duration of the short-term effect is also found in the study by Symons and Robertson (1990) for 16 OECD countries. According to their study, a rise of 1 percentage point in the tax wedge on average leads to an immediate rise in real labor costs by one half of a percentage point, and about half of this effect remains after five years.

1/

The control solution consists of the inflation time paths obtained by keeping the exogenous variables at their end-1995 levels over the forecast period, with the exception of productivity growth, whose growth rate was fixed at its end-1995 value.

1/

Import price changes were introduced as an exogenous variable in the ECM but their effect was found to be insignificant.

1/

With the exception of 1979, when it was just tinder 2 percent.

Finland: Selected Issues and Statistical Appendix
Author: International Monetary Fund