The extended LUCI
(authors: Jan Brůha, Jan Šolc)
The Labour Utilisation Composite Index – LUCI – was introduced two years ago and has now been extended. The labour market plays a significant role in the evolution of domestic inflation pressures, so it is important to know its cyclical position. LUCI makes it possible to aggregate a large number of time series in a transparent manner and hence provides a composite evaluation of labour market tightness. This box presents an extension of the original LUCI concept and outlines its use in short-term forecasting.
Several major improvements have been made as part of the LUCI extension. First, the number of variables included has been increased. Second, the cyclical components of the time series are now filtered in a model-consistent manner, unlike in the previous approach, where they were filtered outside the model. Third, it is now possible to work with time series observed at various (monthly and quarterly) frequencies, which allows for more frequent updates. The lag between LUCI and individual variables can meanwhile be modelled, due to the use of a dynamic factor model instead of the previous static one. This makes it possible to draw up a short-term outlook for the index. The first component of the dynamic factor model, which corresponds to LUCI, explains about 64% of the variability of the cyclical components of the time series included.
The interpretation of historical labour market developments is very similar to that of the original LUCI (see Chart 1). Both the peaks and the troughs of the previous cycles are identified similarly by the two approaches. Their assessment of the currently still exceptionally high labour market tightness is also more or less the same.
Chart 1 (BOX) The original and new definitions of LUCI
The original and new LUCIs assess the labour market cycle very similarly; labour market tightness recently peaked
(index; vertical axis shows standard deviations)
The new LUCI can be used to predict selected labour market and other variables. The ability to predict labour market variables is aided both by the model structure in which LUCI is estimated and by the use of high-frequency (monthly) data. The latter makes the forecast of the overall LUCI – and hence also the subsequent decomposition into individual variables – more accurate. As well as for forecasting, LUCI can be used to evaluate the consistency of the various different labour market time series. In this way, one can, for example, resolve the issue of whether wage growth was consistent with the evolution of other labour market variables. Besides labour market variables, LUCI can be used to prepare a short-term outlook for non-tradables inflation, which LUCI precedes by about nine months (see Chart 2). By contrast, LUCI lags behind the output gap by about two quarters.
Chart 2 (BOX) Correlation between LUCI and selected variables
LUCI precedes non-tradables inflation by approximately nine months; by contrast, it lags behind the output gap
(horizontal axis shows lags in months)
Note: Negative lag values mean that the LUCI precedes the given macroeconomic variable; positive values mean that it lags behind it.
According to LUCI, the current labour market tightness is close to a historical high (see Chart 3). This is due mainly to series linked to unemployment and, to a lesser extent, to wages and costs and other factors, amid roughly equal contributions of the variables aggregated into employment and demand for labour. The contribution of wages and costs is (as in the case of unemployment) at a historical high and is thus consistent with the significant demand pressures currently observed in the domestic economy. The only gradually decreasing labour market tightness in the coming quarters will probably lead to continued brisk wage growth next year.
Chart 3 (BOX) LUCI and its basic decomposition
The current LUCI levels are still very high from the historical perspective, with all components indicating exceptional labour market tightness (index; vertical axis shows standard deviations)
 Previously, cyclical components were isolated using a univariate statistical filter and the principal components method was applied to them. Now the model is formulated in stock form, which allows us to estimate cyclical components using the Kalman filter. This increases the statistical significance of the estimates, as the cyclical components of all series are filtered simultaneously.
 The resulting index is subsequently standardised. This means that for each estimate the long-term mean is zero or, say, a normal labour market situation. Positive values signal higher labour market tightness and negative values lower labour market tightness than the long-term average. The time series are merged into several representative categories for clarity of presentation.