This is the first in a series of blog posts analysing and discussing monthly effective exchange rates time series in the former Yugoslav republics, now four independent states.

Aims

The aims of these blog posts are the following:

• To describe main features of monthly effective exchange rates time series
• To model series using popular time series models
• To forecast future effective exchange rates
• To compare forecasting performance of different time series models

Dataset

The dataset under examination comes from three different sources.

Serbia: The National Bank of Serbia database. It shows nominal and real effective exchange rates (http://nbs.rs/export/sites/default/internet/latinica/80/ino_ekonomski_odnosi/SBEOI11.xls). Both series cover period 2002M1 to 2019M5. Original base was 2005=100, but we have converted these two series to a new base 2010=100, to be comparable with the series for other countries with the same base.

Croatia and Slovenia: Eurostat database (https://ec.europa.eu/eurostat/data/database)

Two monthly series were used: nominal and real effective exchange rate – 42 trading partners (industrial countries). In case of real effective exchange rates the consumer price index was used as a deflator. Both series cover period 1994M1 to 2019M4. Base was 2010=100.

Macedonia: IMF database (http://data.imf.org)

Two monthly series were used: nominal and real effective exchange rates with the consumer price index used as a deflator. Both series cover period 1994M1 to 2019M5. Base was 2010=100.

Dataset description

Dataset is available for download in two formats: csv and RData.

Column name in csv file uses the following format: CccXEERm, where Ccc is the acronym for the country (Cro, Mac, Slo and Ser), X takes values N and R (NEER and REER) and m stands for monthly data. In case of Croatia we have two additional series. The first pair of effective exchange rates (CroNEERm and CroREERm) came from the IMF database, while the second one (CroNEER1m and CroREER1m) came from the Eurostat database. The later pair was used in further analysis.

Those familiar with R could download and use RData file. This file contains time series in different formats, but the multivariate time series object used directly in the next blog posts has the following name: eer.ts. It contains all nominal and real effective exchange rates for four countries. Based on this time series object and few other multivariate time series objects were created and stored in the RData file. These are the objects stored in RData file: a, b, eer.zoo, neer, dneer, reeer, dreer, Cro.err, Mac.err, Ser.eer, Slo.eer, dCro.err, dMac.err, dSer.eer, dSlo.eer. For instance, neer, contains all nominal effective exchange rates for four countries, dneer is the first differences of series in neer. Cro.err contains two series: nominal and real effective exchange rates for Croatia. Period covered: 2002M1 to 2019M4, because Serbian series start from 2002M1 and series for Croatia and Slovenia end in 2019M4.

In the next blog post line graphs for each country monthly effective exchange rates will be presented and discussed.