GDP forecast based on semantic business cycle identification - release: 2008-08-15 |
|||
Special release notes |
|||
| Concept | In this part of the project we develop a so-called surprise index which can be interpreted as a business cycle indicator. We are currently regularly calculating short term GDP forecasts based on this index which has desirable properties due to the exclusiv use of survey data. To date we are in the evaluation phase where we gather experiences with the implementation of the new tool. At the end of this evaluation period (approx. early 2009) an official release will follow that takes into account the experiences made. |
| News | Between the last release (April 2008) and the current release, two major developments affected the estimation of the surprise index. First, there was a major revision of national account data, and second we developed a new version of the indicator that should be easier to communicate to the public. Both changes are discussed below including their implications. |
| GDP data | In July 2008 the Federal Statistical Office (BfS) published revised national account data (quarterly data were provided by the State Secretariat for Economic Affairs, seco). The revised data indicates significantly stronger growth in the past two years compared to previously released figures. |
Indicator values |
1. According to our theoretical concept we use the share of firms that are surprised by the actual economic developments compared to what they previously expected. This surprise can arise from six out of nine situations. Firms may either plan to raise capacity utilisation but are later forced to decrease it. Likewise, no-change / increase; no-change / decrease; raise / decrease, and decrease / raise also indicate surprises. Up until the last release we used a weighted average of all these possibilities for calculating the business cycle indicator. Albeit very successful in terms of tracking and forecasting Swiss GDP the indicator's weights are chosen arbitrarily and are thus potentially difficult to communicate. We have therefore decided to use a simpler version of the indicator by strictly considering only the share of firms who are negatively surprised. More precisely, only those firms which expected to raise capacity utilisation but actually decrease it in the following quarter. 2. We drop the requirement that firms have to answer three quarters in a row for entering the indicator. |
| Effects | The new GDP data implies an upward revision of growth estimates due to higher starting values. Further, the new indicator is less volatile than the original one. A comparison of their most recent values shows that the new definition signals a stable GDP growth rate in the second and third quarter (year-to-year growth) of 2.8% while the original indicator implies a rate of 2.0% in the second and 2.4% in the third quarter. Using the original definition yields a better regression fit (residual standard error: 0.51, compared to 0.58) and slightly better turning point estimates. |
| Conclusions | During the next months we will continue to observe the performance of all potential indicators and decide about the best official release version at the end of the evaluation period. For the time being we publish time series of both indicators. |
| «project home |