Meldung vom 29.01.2018
CEE Phillips Curves: dead or alive?
- Output gaps indicate very clearly that the business cycle has accelerated.
- Beveridge curves indicate tight labor markets in the Czech Republic (CZ) and Hungary (HU).
- Unit Labor Cost growth present clear evidence of wage pressure in Romania (RO), Hungary (HU) and the Czech Republic (CZ) but not in Poland (PL) or the Euro Area (EA).
- This is accompanied by average core inflation being more than double as high in CZ, HU and RO compared to the average rate in PL and EA.
- Estimating CEE Phillips Curves show that the output gap effect is well and active. Though, time-varying estimates indicate a recent weakening.
2018 is projected to be a favorable year for the economies of the CEE region. The business cycle has already accelerated in 2017 with the fastest growth being reported in Romania (8.6 %, y/y, sa) during the third quarter. However, also Poland (5.2 %, y/y, sa, Q3), the Czech Republic (5.0 %) and Hungary (4.1 %) grew well above the Euro Area, which itself is experiencing accelerated economic growth (2.8 %). Hence, convergence has returned to the region.
The acceleration of the business cycle becomes even more evident by looking at output gaps, which reflect the difference between actual and potential GDP as a share of the latter. Potential GDP is the highest level of output which can be sustained over the long term. Figure 1 (see pdf) shows the output gap for the Euro Area and four CEE economies in 2016, 2017 and 2018. While in 2017 the output gap was still negative in Poland (-0.3 %), the Euro Area (-0.1 %) and Hungary (-0.05 %), GDP exceeded its potential in the Czech Republic (+0.5 %) and Romania (+0.5 %). In 2018, the GDP of all five economies will exceed its potential, entering the expansionary phase of a boom and bust cycle. The effect will be most pronounced in Romania.
Labor markets show a similar picture. Short term unemployment rates (unemployed for less than 12 months) are at, or very close to, their historic lows (Figure 2 - see pdf). Job vacancy rates (JVR), which measure the ratio of job vacancies over the sum of occupied and vacant jobs, have increased substantially. A high JVR indicates unmet demand for labor as well as skill mismatches. The Czech Republic shows the tightest labor market with a JVR of 4 % in Q3 2017, followed by Hungary (2.4 %), the Eurozone (1.9 %), Romania (1.2 %) and Poland (1 %).
Relating job vacancies to unemployment rates, gives an indication of the business cycle in the labor market. The negative relationship between vacancies and unemployment is shown by the Beveridge curve. During a cyclical expansion, there should be a high number of vacancies at a low rate of unemployment. Figure 3 (see pdf), shows Beveridge curves for the Euro Area, Poland, Hungary, the Czech Republic and Romania. The Czech Republic, clearly, has the steepest curve and the tightest labor market with the highest JVR and the lowest unemployment rate. The tightness of the labor market is indicated by the position on the curve. For all five economies, the current combination of vacancies and unemployment is the tightest labor market situation since the financial crisis. Compared to the CEE economies it is striking that the Euro Area’s Beveridge curve lies more outward indicating less efficient labor market matching. This outward shift occurred between 2010 and 2014.
The combination of high job vacancy ratios and low unemployment rates is also reflected in workers bargaining power. The growth of hourly unit labor costs, which adjusts wage growth for changes in labor productivity, reflects the CEE Beveridge curves very well. Romania, Hungary and the Czech Republic show the highest wage pressure, while the Euro Area and Poland seem to face less pressing labor shortages (Figure 4 - see pdf).
So far, we have argued that the business cycle has accelerated, labor markets are becoming severely tighter and productivity adjusted wage pressure is emerging rapidly. These dynamics are more advanced in the Czech Republic, Hungary and Romania than in Poland or the Euro Area. The next question must be: What are the implications for inflation?
There have been main concerns among economists, central bankers and financial market observers about the Phillips curve. Particularly, whether the relationship between the state of the domestic economy and price developments has changed dramatically or even ceased to exist 1). Looking at core inflation at constant tax rates, shows that the fundamental macroeconomic differences within our sample of economies is well reflected in inflation dynamics (Figure 5 - see pdf). Core inflation is lowest in the Euro Area (Q4 17: 1 %, y/y) and Poland (1.2 %) where labor market pressure and output dynamics are less advanced. In Romania core inflation accelerated rapidly during the last few months from 0.9 % in the first quarter to 2.2 % in the fourth quarter of 2017. Core inflation in the Czech Republic and Hungary has accelerated more gradually reaching 2.7 % in Hungary and 2.8 % in the Czech Republic in the last quarter of the year.
In spite of the empirical evidence presented, there is no consensus among CEE central banks, whether domestic inflationary pressures are on the rise. The Hungarian National Bank (HNB) decided to loosen monetary policy to further lower the long end of the yield curve. The National Bank of Romania (NBR) has increased its key policy rate for the first time since July 2008 in early January. However, its deputy governor Liviu Voinea was keen to stress that the 25 bp increase should not be interpreted as the start of a rate hiking cycle 2). The Czech National Bank (CNB), on the other hand, attempts to cool down the overheating economy by conducting more restrictive monetary policy. The Polish Central Bank (PCB) is not tempted to act as long as core inflation remains moderate.
The differences among central banks’ perceptions is closely linked to uncertainty regarding the drivers of core inflation. In particular, the uncertainty evolves with respect to the question whether the traditional determinants lost their economic significance. Or put differently, the question is whether the traditional Phillips Curve mechanisms are still active. To shed some light on this question we estimate Phillips Curves for PL, HU, CZ and RO.
The Phillips Curve goes back to a seminal article in 1958 in which A.W. Phillips describes the relationship between unemployment and wage inflation in the UK 3). The Phillips Curve, then, became a vital policymaking tool of the 1960s linking inflation to the unemployment rate and past inflation. During the 1970s when the US experienced a period of high inflation and high unemployment, the Phillips Curve was modified to consider the importance of inflation expectations, known as the expectations-augmented Phillips Curve (Friedman, 1968, and Phelps, 1967) 4). As a response, New Keynesian Macroeconomics combined rational expectations with sticky prices, forming the New Keynesian Phillips Curve (NKPC). The NKPC links inflation to expected future inflation and the amount of slack in the economy. Galí and Gertler (1999) propose a hybrid version of the NKPC which additionally accounts for backward looking price setting behavior 5). Until now, this hybrid NKPC is set as the empirical standard incorporating the following determinants. Inflation in period t is explained by past inflation in period t-s, future expected inflation and a measure of the state of the economy. Economic slack is most often proxied by the output gap. Alternatively, measures of labor market slack related to the unemployment rate are used. With the increasing internationalization of the global economy, non-domestic price setting might have gained importance for domestic inflation dynamics. Particularly so, for small open economies like those of the CEE region 6). Hence, we add a term measuring imported inflation. Furthermore, instead of adding a constant we use the inflation target. Our Phillips Curve takes the following form (see pdf).
Estimating the coefficients of our Phillips Curve for the period from Q1 2002 to Q3 2017 yields remarkably similar results for the domestic output gap coefficients. A 1%-point higher output gap, which is constructed by applying a Hodrick-Perscott-Filter, is associated with an average 0.35 %-points higher inflation rate. The respective inflation rate is annualized quarter-on-quarter seasonally adjusted core inflation, HICP excluding energy and unprocessed food, at constant taxes. Domestic price pressure emerging from the business cycle are well and active.
Table 1: Phillips Curve estimates from OLS regressions (see pdf)
Imported inflation, which we measure based on price developments of imported goods, is an important additional mechanism. Particularly in CZ and PL, where a 10%-point increase in the inflation rate of imported goods is associated with a 0.3 %-points higher core inflation. Hungary shows a similar coefficient, though, due to higher variability it cannot be differentiated from zero at a statistically significant margin. In Romania imported inflation seems to play a less prominent role. Standardizing the coefficients from table 1 by their respective standard deviation, also known as beta coefficients, shows that the relative importance of domestic price pressures compared to imported inflation is strongest in the Czech Republic. In Poland both mechanisms are of a comparable economic magnitude. The results show that both mechanisms are at play. However, it should not be argued that imported inflation dominates domestic price pressures.
The Phillips Curve estimates show that over the period Q1 2002 to Q3 2017, the state of the economy does have a significant effect on inflation, particularly in the Czech Republic and Poland. It has, however, been argued that the Great Recession has structurally changed the Phillips Curve 7). Hence, it would be interesting to know to what extent the coefficient estimates of table 1 have changed over time. This exercise can be conducted by rolling the observation window of the regressions, resulting in time-varying coefficient estimates 8). Figure 6 (see pdf) shows the results for the output gap coefficient θ. There is clear evidence that the influence of the output gap on core inflation has become weaker during the most recent period. Poland and Romania have experienced a relatively sharp drop in the output gap coefficient in 2016, while the decline was more continuous and started earlier in the Czech Republic and Hungary.
The analysis has shown that the business cycle in the CEE region has accelerated. This can be seen not only in GDP figures but also in the labor market. In line with the Phillips Curve, domestic price pressure is emerging at an accelerated pace. Estimating hybrid New Keynesian Phillips Curves support this descriptive observation. However, looking at changes over time shows that the relationship between the state of the domestic economy and domestic inflation developments has weakened. Local central banks need to observe this development very carefully for successfully conducting monetary policy.
Martin Ertl Franz Zobl
Chief Economist Economist
UNIQA Capital Markets GmbH UNIQA Capital Markets GmbH
This publication is neither a marketing document nor a financial analysis. It merely contains information on general economic data. Despite thorough research and the use of reliable data sources, we cannot be held responsible for the completeness, correctness, currentness or accuracy of the data provided in this publication.
Our analyses are based on public Information, which we consider to be reliable. However, we cannot provide a guarantee that the information is complete or accurate. We reserve the right to change our stated opinion at any time and without prior notice. The provided information in the present publication is not to be understood or used as a recommendation to purchase or sell a financial instrument or alternatively as an invitation to propose an offer. This publication should only be used for information purposes. It cannot replace a bespoke advisory service to an investor based on his / her individual circumstances such as risk appetite, knowledge and experience with financial instruments, investment targets and financial status. The present publication contains short-term market forecasts. Past performance is not a reliable indication for future performance.
1) See Gordon, R., (2013), The Phillips Curve is Alive and Well. Inflation and the NAIRU during the Slow Recovery, NBER Working Paper 19390, National Bureau of Economic Research.
2) Comment at the Euromoney Central & Eastern European Forum 2018, Vienna, 16th – 17th January 2018.
3) Phillips, A.W., (1958), The relationship between unemployment and the rate of change of money wage rates in the United Kingdom, 1861-1957, Economica, 25(100), pp- 821-852.
4) Friedman, M, (1968), The Role of Monetary Policy, American Economic Review, 58, pp. 1-17. Phelps, E.S., (1967), Phillips Curves, Expectations of Inflation and Optimal Unemployment over Time, Economica, 34, pp. 254-82.
5) Galí, J., and Gertler, M., (1999), Inflation dynamics. A structural econometric analysis, Journal of Monetary Economics, 44(2), pp. 195-222
6) See Szafranek, K., (2016), Linking excessive disinflation and output movements in an emerging, small open economy. A hybrid New Keynesian Phillips Curve perspective, NBP Working Paper No. 239, Economic Institute Warsaw.
7) he Hungarian Central Bank, for instance, states in a special topic of its September 2017 Inflation Report that “the role of external factors in domestic inflation developments strengthened in the past period, and after 2012, the changes in inflation in Hungary were mainly influenced by global effects.” (p. 64).
8) We hold the sample size constant at 36 quarters.