ANNEX Data Sources and Variable Construction
Unit labor costs in domestic currency for industry i in country c is calculated as:
The relative unit labor costs between Ireland and country c in industry i are given by:
where ec is the nominal exchange rate of country c relative to the Irish currency.
Effective unit labor costs in industry i are constructed by summing relative unit labor costs of Ireland over all its trading partners, where the weights, twci, are based on the shares of Irish exports in industry i that are destined to each country.
An aggregate index, corresponding to a real effective exchange rate (REER), is then constructed by summing over all industries, using weights, ωi, that can either depend on the employment share of each industry or the output share of each industry in total manufacturing.
The data frequency for all variables is quarterly.
The Central Statistical Office (CSO) of Ireland was the source of Irish data on production, employment, hours worked, and wages by NACE sector.
OECD was the source of partner country data on unit labor costs. In particular, data on production and employment was available from the OECD Structural Analysis (STAN) database by NACE industry for each country. Hours worked per employee in the business sector and hourly earnings of the manufacturing sector were taken from the OECD Analytical Database for each country on an aggregate basis. Missing observations, which were more frequent at the end of the sample, were supplemented by unit labor costs in domestic currency for the business sector, from the OECD Analytical Database. Nominal exchange rates were obtained from the IMF’s International Financial Statistics.
Irish exports to each country by each industry were taken from the World Bank’s World Integrated Trade Solutions, which contains trade statistics from the COMTRADE database made available by the United Nations Statistics Division. SITC industry classifications were matched by description to NACE industries to obtain sectoral direction of trade estimates. Export weights were based on averages over 1998–2000.
The CSO was the source of data on the employment share of each industry in total manufacturing in Ireland and 1995 value-added shares of each industry. The weights used to construct the output-weighted REERs were based on output shares at the beginning of 2002, which were obtained by accumulating 1995 value-added shares of each industry using production indices.
Final index calculations were seasonally adjusted and rebased to 1995: q3 = 100.
Prepared by Valerie Cerra and Jarkko Soikkeli.
Throughout this paper, a decline in a REER or NEER index devotes a depreciation or gain in competitiveness. See Annex for data sources and variable definitions.
The developments across sectors were, however, uneven and may be partly related to very strong productivity gains that are characteristic of cyclical turning points. Therefore, it could be premature to suggest that the Irish manufacturing productivity has returned to its earlier trend.
The real effective exchange rate as reported in the IMF’s International Financial Statistics (IFS) matches the index of output-weighted REER relatively closely. However, the recent pick-up in the latter index goes largely unnoticed in the IFS index because it is smoothed using a Hodrick-Prescott (H-P) filter. There are tradeoffs between these measures. The filtering used in the IFS index smooths out cyclical swings in productivity to obtain a gauge of underlying trends. On the other hand, H-P filters suffer from well-known end-point problems. Moreover, it can be argued that the output-weighted index measures competitiveness more accurately than the IFS index, since it applies industry-specific partner-country trade weights, rather than partner weights based on aggregate export trade.