A systems approach to the determination of the NAIRU, inflation and potential output in Austria. Friedrich Fritzer and Heinz Glück 1 - PDF

A systems approach to the determination of the NAIRU, inflation and potential output in Austria Friedrich Fritzer and Heinz Glück 1 Introduction After decades of very low unemployment rates, tensions in

Please download to get full document.

View again

of 12
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.

Devices & Hardware

Publish on:

Views: 18 | Pages: 12

Extension: PDF | Download: 0

A systems approach to the determination of the NAIRU, inflation and potential output in Austria Friedrich Fritzer and Heinz Glück 1 Introduction After decades of very low unemployment rates, tensions in the labour market also appeared in Austria by about the middle of the 1980s. Under the influence of the opening of the Eastern European countries, Austria's accession to the EU, globalisation and other shocks, the discussion has intensified whether and to what extent these developments may have contributed to the ongoing rise of unemployment and whether they will continue to do so. Problems of this kind are frequently discussed in terms of the natural rate of unemployment or the NAIRU (non acceleratinginflation rate of unemployment), which relate labour market concepts to inflation and potential output. As is well known, the NAIRU seems have risen in the last few years - not only in Austria, but in many other countries. From a policy perspective, it is of foremost interest whether the NAIRU can be sustainably reduced through deliberate economic policy measures by working, for instance, directly through labour market institutions and the wage bargaining process (OECD 1994). However, Ball (1996) pointed out that the disinflation policy of the last years has caused a considerable outward-shift of the NAIRU in many countries (i.e. the stability of the price level is now associated with higher unemployment than before). During the 1970s, Austria's unemployment rate fluctuated within narrow margins around 2%. In 1980, unemployment started to accelerate, reaching 6.6% in Immediate questions are: what were the reasons for this development and what can we do about it? In case of Keynesian unemployment, we should call for measures to stimulate effective demand; i.e. higher public spending or a looser monetary policy. At the moment, these measures are not very popular because of the binding constraints of the Maastricht criteria; and if we scan the data we cannot actually find any indication that the rise in unemployment is linked to too restrictive a course in terms of public spending or monetary policy. Budget deficits have been rising since the 1980s, and the downward trend in inflation already started in the mid-1970s when unemployment was pretty stable at about 2%. Classical unemployment would instead call for measures to reduce the rigidity of real wages. In the analysis below, we will not deal with this case because the wage bargaining process among the social partners in Austria is meant to prevent unemployment. Other sources of unemployment may be changes in production patterns which affect the demand and supply side of the labour market. Demographic developments, generous unemployment benefits, non-wage labour costs are some of the factors which drive the equilibrium unemployment rate. Unfavourable shifts in these determinants will increase the natural rate of unemployment and induce a parallel movement of actual unemployment or an ever accelerating inflation rate. If this is the case, the only wise response to high unemployment lies in the removal of structural impediments to the way the labour market works. This paper tackles the questions raised above. After a short review of the results of some recent studies on Austria in Section 1, we specify and estimate single equations for wages, prices, 1 The authors are economists at the Oesterreichische Nationalbank, Economic Analysis Division. The views expressed in this paper are those of the authors and not necessarily of the institution with which they are affiliated. 115 output and unemployment in Section 2. In Section 3, this system of equations is estimated by 3SLS, and in the final section we try to draw some conclusions from the results. 1. The discussion in Austria This section gives a short - and certainly incomplete - review of the discussion on wageprice relations, the Phillips curve, potential output and the NAIRU in Austria. As far as the more recent investigations of these issues in Austria are concerned, we start with Wörgötter's (1983) contribution, in which he tries ... to give an explanation for the astonishing stability of the relationship between wage and price increases on the one side and unemployment rates on the other side in Austria from the end of the post-war recovery in Contrary to the experience in other industrialised countries no outward shift of the Phillips curve could be observed during the 1970's... . Wörgötter develops an equation system for the simultaneous determination of wages and prices and concludes from the results that the ... influence of social partners on the determination of wages and prices in Austria allows for a stable long-run empirical relationship between inflation and unemployment... The self-adjusting behaviour of social partners in Austria...remo ves this constraint on economic policy which can rely on the endogenous stabilisation of wages and prices, without being forced to apply an anti-inflationary fiscal and monetary policy. In the context of monetary and exchange rate policy, Handler (1989) took a somewhat different view. He feels that, due to external influences (price shocks), the Phillips curve for Austria cannot be interpreted as a trade-off between inflation and unemployment. Therefore, monetary policy decisions should be only loosely based on this relation. Moreover, he did not find any empirical proof that the permanent dampening of the inflation rate through the hard-currency option had negative effects on the real economy and on unemployment. Thus the Phillips curve is no constraint for a stability-oriented monetary policy. Pichelmann (1989), however, showed that the NAIRU doubled during the fifteen years preceding his investigation. He interprets this rise as evidence of ... an increase of distributional pressures in the Austrian economy and addresses the question whether the Austrian political and organisational system may have lost some of its flexibility. Employing a simple macro-economic model of wage-price formation, he establishes that both the increase in employers' contributions to social security and the rising share of long-term unemployment associated with higher overall unemployment have exerted pressure on real product wages and thus exerted upward pressure on the equilibrium rate of unemployment . Though hysteresis will have played an important role, the rise of NAIRU was regarded as evident. Wehinger (1990) pointed out the problem that potential output and the natural rate of unemployment are frequently determined independently or only in a recursive way, and successfully tries a simultaneous approach. He identifies some marked output gaps between 1967 and 1988, but finds that they are slightly smoother compared to calculations using simpler methods. Gartner's study (1995) represents a first attempt by Oesterreichische Nationalbank to introduce the concepts of potential output and the output gap into monetary analysis. Trend output was determined by means of the Hodrick-Prescott filter, and Granger causality tests were used to check whether the values for the output gap could contribute to forecasting inflation, which was confirmed. Hahn and Rimstler (1996) try to combine astructural (e.g. the Hodrick-Prescott filter) and structural approaches in determining potential output, the latter being deduced from wage-price equations. They find that actual GDP in Austria was almost always below potential output since the beginning of the 1990s, and that the average length of cycles was 28 quarters, with the cyclical component of unemployment lagging behind the GDP cycle by about three quarters. 116 Pichelmann and Schuh (1996) point out some problems inherent in the concept of NAIRU. They survey a number of hysteresis-mechanisms which could lead to permanent shifts of equilibrium unemployment over time, implying that a long-run NAIRU may not even exist. Other problems are that empirically derived NAIRU estimates are highly dependent on model specifications, and that a considerable amount of statistical imprecision is inherent in the estimates. Therefore, they argue, any policy conclusions and recommendations drawn from the NAIRU must be judged with utmost care. 2. Single equation results With these caveats in mind, we approach the task of finding some empirical evidence for the items involved. As mentioned, it is necessary to simultaneously estimate an equation system like the one below 2 in order to get consistent parameter estimates for the determination of wages, prices, potential output, and unemployment: v =a 0 +a x p c +a 2 q pot +e (û (1) Aw = ß,Ap c + ß 2 4pl 1 + p3a(c/ - U nat ) + ß 4 AA( y- U ) + ß 5 A(/-p) + ß 6 A(/ - p) ^ + ß ^ ^ + e Ao) (2) AC/ = YjA^ - y pot ) + y^') + Y 3 A(Aapp+ Aapp^) / 2 + y 4 A«w/c + y 5 Aunion + e Au (3) Ap = 8q + Sj ( Aule + Aulc_\ + Aulc_2 ) / 3 + Ô2Ap m + (4) /* k = ^+ + Ç 2 (y - y pot ) + (5) Equation (1) reflects the target-real-wage-bargaining model. In such an environment, unemployment in excess of the natural rate can coexist with stable wage and price inflation. Equation (2) is an equation for the short-run dynamics of wage inflation. The growth of wages Aw relates to consumer price inflation Ap c, the change in the unemployment gap A(\J-U nat ), the growth of the gap between consumption and output prices A(p c -p) and the lagged residuals from equation (1), which can be interpreted as the deviations of real wages from their target level. The change of the unemployment rate (equation (3)) is determined by the deviation of real output from its potential value y 0 ', 3 the growth of non-wage labour costs Anwlc, the growth of union density Aunion and the growth of the share of apprentices as a percentage of the labour force Aapp. The evolution of prices (equation (4)) is estimated from unit labour costs Aule (the divergence between the growth rates of wages and potential productivity) and from the growth rate of import prices Ap m. The production function (equation (5)) is solved for multi-factor productivity {y-q.15h-q.25k) which is determined by a time trend T and the output gap. The preliminary specification search has to be performed by single-equation estimates using OLS. This is done in the following paragraphs. 2 This approach goes back to Coe and Krueger (1990), who apply the target real wage bargaining model to the German labour market. 3 In the single equation estimation all potential values have been proxied by a Hodrick-Prescott filter. 117 2.1 The wage equation Starting with the general wage formulation of equation (2), we tried to explore some additional and specific issues, which did, however, they did not contribute very much to the overall picture. We tried to include variables or proxies for unionisation, formulated more complex hypotheses for price expectations, and also experimented with error correction terms. Finally, we ended up with an estimation of the following form: Aw = 0.882V Ap!i a(í/ - U ') AA(í7 - U ') a ( / - p) (5.39) (4.35) (-4.05) (3.59) (-3.88) â(/-p)_ i e _,+e a ( ^ (-3.02) (-2.24) Numbers in parentheses are t-values. R 2 = D.W. = Estimation Period: This result stands very much in the tradition of the Austrian wage-formation process, showing almost full compensation for the inflation of the past year (which could also be interpreted as a simple expectation formation). The error correction term of the level equation (1) is significant and correctly signed. 2.2 The unemployment equation Estimation of equation (3) gives the following result: AU = a(j; - ) a(7 - y po ') A{Aapp+ Aapp^) / 2 (-5.64) (-2.57) (-3.91) +0.0\6 Anwlc+ Q.076Aunion + Au (3') (3.62) (2.60) Numbers in parentheses are t-values. R 2 = D.W. = Estimation Period: This result conforms reasonably well to the expectations. Positive deviations of real output from its trend development reduces unemployment to a rather modest extent. As observed in many other studies, rising non-wage costs foster unemployment. 2.3 The price equation According to equation (4), the evolution of prices is determined by the relation between wages and productivity and foreign prices. We get the following result: Ap = (Am/c + A w / c l, + A w / c ^ / S + O.OSOV + 6 ^ (4 1 ) (2.131) (5.625) (1.79) Numbers in parentheses are t-values. R 2 = D.W. = Estimation Period: Thus, price developments are mainly determined by changes in unit labour costs. 2.4 The production function The production function is, as mentioned, solved for productivity, and estimation of equation (5) exhibits the following: 118 j Ä k = r (7 f - ' ) +e, (5') (772.17) (42.53) (5.11) Numbers in parentheses are t-values. R 2 = D.W. = Estimation Period: output gap. Accordingly, overall productivity is determined by an autonomous time trend and by the 3. Results of the system estimation and interpretation A system approach has some advantages over a partial equilibrium approach when estimating the natural rate of unemployment. In particular, the labour market gap will not only have an affect on wage inflation, nor will the product market gap have an affect on price inflation alone. Instead, the parameters that define the natural rate of unemployment and potential output are themselves estimated jointly with the parameters in the wage and price equations. The specifications for the system are those of the single equation estimation and are repeated here for the sake of convenience: C pot w =a 0 + OCj/J + a 2 q + m (6) Aw = V l Ap c + + ß3A(c/ - U at ) + ß 4 AA(c/ - C/ a ' ) + ß 5 A(/ - p) + ß 6 A(/ - p) ^ +ß.e^ + e At0 (7) AU = y x A{y-y pot ) + y 2 A{y-y pot ) + y 3 A{Aapp+ Aapp_^) / 2 + Y 4 Amv/c + y 5 Aunion + Am (8) Ap = ô 0 + ôj ( Aule + Am/c.j + AUIC_2 ) / 3 + Ò2Ap m +8^ (9) /* * = C 0 + ÇjT- + Ç 2 (y - y pot ) + e y (10) where U at = ^Á^PPk + Acipp ^ ) / 2 + Y4 ^Anwlc k +y 5 ^ Aunion k (11) *=71 *=71 *=71 y po ' = 0.75h pot k pot + Ç 0 + (12) / o i =-0.25h pot +0.25k pot +Co +^T + l po1 (13) Ay pot - Ah pot + Al pot EE Aq pot = -0.25Ah po ' Ak po1 +ÇJ + Al po ' (14) Compared with the single-equation estimates, no proxies for potential output or trend labour productivity plus the natural rate of unemployment are used. Instead it is necessary to substitute expressions (11)-(14) into equations (6)-(10). The constant 1.42 in equation (11) is the unemployment rate in August 1971, which was the lowest rate during the sample period. The determinants of potential output have been expressed at their potential levels. For labour input (h), this is done by calculating total man-hours based on trend 119 average hours per employee, and employment levels consistent with unemployment at its natural rate. The stock of capital is corrected by a capacity utilisation rate. 4 I is the log of the annual working hours per employee. The system has been estimated using non-linear three-stage least squares. Endogenous variables in the system are the contemporaneous values of the dependent variables as well as the growth of labour productivity. The estimation period is from 1970 to Table 1 System estimates Equation w AU Aw Ap J-0.75A-0.25Ä «0= i = o= Ço=3.511 (-2.51) (-2.48) (6.59) (5.11) (521.86) a, = = ß2= , = CO O ö II (17.07) (-5.12) (2.28) (5.22) (26.51) *2= =0.903 ß3 = = ^2 = (2.83) (-2.94) (-3.58) (2.70) (5.31) 0*3 = =0.016 ß 5 = (-3.669) (2.66) (-7.38) 7 5 =0.065 ß 6 = (2.201) (-2.92) ß 7 = (-1.90) R 2 = R 2 = R 2 = R 2 = R 2 = R 2 = R 2 = R 2 = R 2 = R 2 = SEE= SEE= SEE= SEE= SEE =0.009 The system estimation results imply the following expressions for the natural rate of unemployment and potential output: T T T U at = X(Aappk + topp-i) / Mwlc k Y,Aunion k (15) k=l\ k=l\ k=l\ y pot =0J5h pot +0.25k pot (16) Estimates of the historical path of potential output and output as well as the output gap and the growth rates of potential output are shown in Figures 1, 2 and 3. One feature of the results is that the output gap fluctuated during the 1970s with spikes in 1974, 1977 and Subsequently the output gap decreased with a tendency to increase after 1988 (due to the lack of capital stock data, potential output could not be estimated for the years after 1992). 4 In a first approximation the August unemployment rate is considered to be the natural unemployment rate. In the yearly cycle the August unemployment rate is usually the lowest one. Total employees plus the fraction of the unemployed not considered as structural is the potential supply of employees (EP 0 '). Potential hours per employee (LP 01 ) is calculated as the long run equilibrium of an OLS regression of hours per employee on a constant, on trend and lagged hours per employee: (L=f(l,T,L_]). Potential labour input is hence W'^LP 1 EP' . A capacity utilisation index is constructed via yearly survey results of the Konjunkturtest (KT) and the Investitionstest (IT). An OLS regression of (100-IT) on a constant and KT yields the capacity utilisation variable with which the private capital stock is adjusted to yield potential private capital. Public sector real capital is assumed to be fully employed. 120 Figure 1 Output and potential output (in 1,000 billions of Austrian schillings) YPOT Year 4 Figure 2 Output gap (as a percentage of potential output) Year 3.0 Figure 3 Potential output (annual percentage changes) Year 121 Estimates of potential output along with the contributions to growth of changes in potential factor inputs and multifactor productivity are shown in Table 2. Table 2 Contributions to the growth of potential output Annual percentage changes Potential output Labour Capital Multifactor productivity Slower growth in factor inputs contributed most to the slowdown in the growth of potential output during In the period labour growth picked up whereas capital did not return to its contribution of the 1970s. Multifactor productivity gradually decreased over the whole estimation period. The natural rate of unemployment is estimated to have increased steadily from a low of 1.5% in 1971 to a high of 6.3% in The increase in the natural rate of unemployment gained momentum particularly during the 1980s. The unemployment gap widened during the 1980s but narrowed in the last couple of years. Mirroring the development of potential output, the same three periods seem to have crystallised, with different trends in actual and natural unemployment - pretty stable development during , a drastic pickup after 1980 with a narrowing of the unemployment gap after According to the estimation, the bulk of the increase in unemployment after 1980 can be attributed to a rising natural unemployment rate. 7 6 Figure 4 Actual and natural unemployment rates UNAT Year 122 Table 3 Contributions to changes in the natural rate of unemployment Changes in actual unemployment rate Changes in natural rate Due to non wage labour cost Due to changes in apprentices Due to changes in union density Conclusions What lesson may we draw from this exercise? The rise in unemployment after 1980 seems to be determined by structural factors which raised the natural unemployment rate. We estimated a steady increase of the NAIRU from 1.5% in 1971 to 6.3% in The rise gained momentum particularly in the 1980s and the gap between the NAIRU and actual unemployment widened in this period, but has narrowed again recently. From this point of view, measures to remove the structural impediments responsible for the malfunctioning of the labour market are called for. We have identified three microeconomic factors, namely non-wage labour cost union density and the share of apprentices in the labour force. One policy conclusion from this result would be that measures should be taken to speed up an easier integration of young people into the Austrian labour market and the flexibility of the labour market should be enhanced - for example with respect to aspects of labour market regulation which primarily affect the cost of labour; e.g. labour cost advantages of employing part-t
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks