| Summary: | In this paper we study volatility of national accounting aggregates for Iceland and compare it to volatility of these aggregates for other OECD countries. The paper uses three different methods to measure volatility: 1) log deviations from trend obtained using HP filter; 2) log deviations from trend obtained using the filter suggested by J.D. Hamilton; 3) log changes in the series. The paper studies effects of filters like seasonal adjustments on measured volatility. In most cases seasonal factors account for much of the variance in the unadjusted time series. Iceland and Ireland are outliers in this respect asseasonal variations explain relatively small part of the variance in the unadjusted series. We compare volatility of quarterly data to volatility of annual data and derive approximate formulas for the measures of volatility in annual data in terms of voltility and autocorrelations of the quarterly series. In most cases measured volatility in quarterly and annual data give similar pictures of the volatility of national accounting aggregates in a given country, but there are exceptions. Iceland is an outlier in this respect as the increase in the volatility of annual data for consumption is very large compared to volatility of seasonally adjusted data when log changes in the series are used to measure volatility. Much higher autocorrelations in the data forseasonally adjusted consumption compared to GDP explain why consumption can be less volatile than GDP in terms of seasonally adjusted quarterly data, but much more volatile in terms of annual data. We also study the relationship between the volatility of consumption and volatility of GDP. In almost half of the countries in our data set of 34 OECD countries consumption is more volatile than income measured by GDP. Finally, this paper studies the relationship between the size of an economy and its volatility and use it to assess if the Icelandic economy is more volatile than is to be expected when its size is taken into account.
|
|---|