A Nowcasting Model for GDP and its Components

18/01/2011 |  Kriaf Tomer
All Press Releases In Subject:
The Economy and Economic Activity
Abstract
Monetary policy operates in an environment of great uncertainty about the current economic situation. A wide range of economic indicators available in real time are used, therefore, to shed light on various sectors of the economy. This paper presents a method for weighting the indicators to obtain a macroeconomic picture in a resources-uses framework, by using a nowcasting model for the rate of growth of uses, imports and GDP in Israel. The use of the model enables an estimate of economic activity to be obtained one or two months prior to the official publication of the Central Bureau of Statistics. The method incorporates: a switch from a monthly data set that includes full or partial information on the quarter under review, to a quarterly system; a forecast of each of the components of the National Accounts (private consumption, investment, exports and imports); and an estimate of GDP using the sources-uses identity and complementary indicators.
The main findings of this study are:
(1) The indicators that contribute most to forecasting ability are the foreign trade data: goods imports and exports, imports of durables and imports of capital goods; data from surveys--the Consumers Confidence Index, the Purchasing Managers Index, the Capital Utilization Index; and share price indices and data on tax revenues. The advantage of most of these indicators is their relatively low rate of revision, especially at turning points.
(2) It was found that the model functioned well in out-of-sample forecasting of GDP, private consumption and imports, and to a lesser extent, of exports. The quality of the forecast of fixed investment, however, was low, and public consumption could not be forecast at all.
(3) The estimate of the GDP growth rate had an average error (absolute value) of 2.6 percentage points, in annual terms. The model's good forecasting ability of GDP derives from, among other things, the sources-uses identity, which means that errors in uses are offset by imports.
(4) It was found that imports of durable goods are a central indicator in forecasting private consumption, even though they constitute less than 10 percent of consumption.
(5) The estimate of exports yielded the result that the US Purchasing Managers Index can serve as an index of global demand, and is preferable to other well-used international indicators such as world trade and GDP, which become available only after a considerable delay.
(6) The low forecasting ability of fixed investment derives from the difference between the development in residential and nonresidential investment.
(7) The data on investment in inventory are very volatile and subject to frequent revisions, but nonetheless evidence was found that stocks serve as a buffer at times of temporary gaps between supply and demand.
(8) Experience gained so far from using the model shows that its nowcasting errors on real time are not bigger than the out-of-sample errors. Despite the many revisions that the monthly indicators undergo. This is apparently due to the high correlation between the revisions to the indicators for forecasting uses and the indicators for forecasting imports.
 

The full article in English - PDF file  
The full article in Hebrew - PDF file