Search-and-matching frictions and labor market dynamics in Latvia

This paper examines, in an estimated, full-fledged New Keynesian DSGE model with Nash wage bargaining, sticky wage and high value of leisure akin to Christiano, Trabandt and Walentin (2011), whether search-and-matching frictions in labor market can explain aggregate labor market dynamics in Latvia. If vacancies are not observed, the model can, to a reasonable degree, generate realistic variance and dynamics of unemployment, and the correlation between unemployment and (latent) vacancies, but at the expense of too volatile vacancies. As a by-product, one-quarter ahead forecasts of hours worked and GDP exhibit less excess volatility and thus are more precise, compared to a model without search-and-matching frictions. However, if both unemployment and vacancies are observed and a shock to the matching efficiency is allowed for, then the cyclical behavior of forecasted vacancies - and the correlation between unemployment and vacancies - tends to counter the data (to the benefit of better fit of vacancies’ volatility), and the smoothed matching efficiency is counter-intuitively counter-cyclical. Hence the model cannot fit the three statistics - variance of unemployment and vacancies, and the correlation between the two - simultaneously.