An Open Economy DSGE Model with Search-and-Matching Frictions: The Case of Hungary

ABSTRACT

This article builds and estimates a medium scale, small open economy DSGE model augmented with search-and-matching frictions in the labor market, and different wage setting behavior in new and existing jobs. The model is estimated using Hungarian data between 2001–2008. We find that: (i) the inclusion of matching frictions significantly improves the model’s empirical fit; (ii) the extent of new hires wage rigidity is quantitatively important for key macro variables; (iii) labor market shocks do not play an important role in inflation dynamics, but the structure of the labor market influences the monetary transmission mechanism.

KEY WORDS:

• DSGE
• wage rigidity
• Bayesian estimation
• Hungary

Notes

1. Pissarides (2009) and Haefke, Sonntag and van Rens (2013) find that wages of new hires are flexible. On the other hand, Bewley (1999) and Gertler, Sala and Trigari (2008) argue that fairness and motivation lead firms to set new hires wages relative to existing jobs. Survey evidence of the Eurosystem Wage Dynamics Network also supports the latter view (see Galuscak et al. 2012 for cross-country evidence; and Kézdi and Kónya 2011 for Hungary).

2. See Berument, Dogan and Tansel (2009) for evidence in Turkey along similar lines.

3. Benczúr and Rátfai (2014) for a comprehensive description of business cycle facts in many countries, including Hungary.

4. For the sake of brevity, we report only the most important aspects of the model. A detailed description is given in an unpublished Appendix, available from the authors upon request.

5. Christoffel, Kuester, and Linzert (2006) assume that vacancy costs are given by $\kappa /{\lambda }_t$, which makes these costs procyclical. On the other hand, Yashiv (2006) and Fujita and Ramey (2007) argue for countercyclical vacancy costs, since these amplify unemployment fluctuations. Since we include wage rigidity and a large set of shocks, and hence do not need extra amplification, we opt for the simplest case of constant vacancy costs.

6. In what follows we use variables with a hat to indicate log deviations from the steady state.

7. This is the point made by Pissarides (2009). “In the search and matching model, the timing of wage payments during the job’s tenure is not important for job creation. Job creation is driven by the difference between the expected productivity and the expected cost of labor in new matches. … as long as the firm and the worker use the Nash wage rule to split rents at the time of job creation, the job creation conditions are unaffected by the rule used to split rents in ongoing jobs. So wages in continuing jobs may be completely fixed, and yet, if wages in new matches satisfy the Nash wage equation, the volatility of job creation will be unaffected by their wage stickiness.” (p. 1340).

8. See Yilmazkuday (2008) for evidence on structural breaks in transition economies.

9. OECD, Benefits and wages: tax-benefit indicators (2007).

10. This holds also for the other common parameters that we did not include in the table for brevity. The two parameters specific to the EHL mode, ${\sigma }_w$ and φ. are estimated, but the posteriors basically mimic the priors with means 6 and 1, respectively.

11. In the discussion that follows we sometimes use the abbreviation NHWR for brevity.

12. In response to a technology shock the differences are very small (not reported). The UIP shock IRFs are very similar to monetary policy shocks, i.e, NHWR is important for employment only (not reported).

13. Note that the real wage IRFs are very similar throughout, which should not be a surprise. Job destruction and job creation are low in Hungary, so average wage inflation is mostly determined by the evolution and rigidity of wages in existing jobs.

14. Another difference is that we focus on employment instead of unemployment. If labor force participation (not in our model) affects these differently, the measured role of inflows and outflows the two-state model might also be different. See Elsby, Hobijn and Sahin (2013b) for some new evidence on the participation margin, and Campolmi and Gnocchi (2011) for putting participation into a DSGE model.

15. We omit the role of initial conditions on the figure, It can be computed by subtracting the sum of the shoek contributions from the total.