The European Business Cycle Clock 

 

1. Introduction

 

The Business Cycle Clock (BCC) of Eurostat is a new graphical interface[1] for displaying developments of economic key indicators within Europe. It helps users in their understanding of economic up- and downswings, and of the evolution in time and across countries. It is a visual aid as well as a portal towards European key statistics.

2. How does it work?

 

The BCC covers a wide field of economic indicators, such as GDP, consumption, investment, exports and imports, (un-)employment, Economic Sentiment, industrial production, producers prices, production in construction, deflated retail sales. Each of these indicators has been re-calculated in order to reveal more clearly their multi-yearly cyclical development. Short-term increases and decreases due to, for instance, seasonality have been removed from the series. The various steps of calculation are described in more detail in section 5.1.

 

The cyclical behaviour displayed by the graph (see figure 1) distinguishes four phases: (1) the indicator is above its long-term trend and is increasing, (2) it is still above the long-term trend, but after having reached a peak it is now gradually moving downwards, (3) after subsequent decreases it reaches levels below the long-term trend and is heading into a trough, and (4) gradually picks up again by pulling itself out of this trough with positive numbers even though still under the long-term trend. Each of these phases are coloured correspondingly. In general the indicators move in time like the arrows, evolving from one phase into the other. Sometimes a full cycle takes around 4 years, other times it could be more or it could be less; there is no rule of thumb for forecasting cyclical time span.

 

Figure 1. Four basic phases of cyclical development  

 

Besides the graph, users have an overview of the most recent (growth) rates of the indicators for the countries which have been selected (see figure 2).

 

Figure 2. Selection of countries and indicators, and their latest values

 

In here users can also select the indicators they want to include in the graph by checking the boxes. 

 

The BCC combines both graph (figure 1) and selection pane (figure 2) into a one page interface including an animation control below the graph for watching economic evolution in a dynamic fashion. As can be seen from figure 3 each indicator is represented by its own graphical symbol, and by mouse-over the symbol the title of the indicator is highlighted within the graph. In this example the German economic picture in August 2006 looks quite positive as almost all indicators are clustered well above their long term trend average and are increasingly doing so. By running the animation a user can watch all subsequent stages of the business cycle (from 1990 onwards). These stages evolve gradually as some indicators can be leading (e.g. Economic Sentiment and exports), and some can be lagging (e.g. private consumption). With the tabs on top of the graph the user can immediately switch to France or see both French and German indicators simultaneously in one graph for cross-country comparisons. A user can also save their selection (of indicators and countries) through the bookmark button (in the top right) in their Favourites or for mailing purposes. More detailed information on each indicator can be activated through the four buttons to the right of each title, opening: a short description; a table with more data, and customizable graphs and maps.

An example (GDP) of these tables, graphs and maps can be found in the annex.

 

Figure 3. Business Cycle Clock for the selected countries Germany and France

3. Current status of the BCC and next steps of improvement

 

Eurostat is developing its new web interfaces such as the BCC in an incremental way, starting with basic functions – i.e. the current BCC- gradually adding new functions and improving the look and feel.

 

In the next BCC version users will be able to display more or all countries in a single graph for one or a limited number of indicators, thereby extending their view on possible cross-country dependencies. Also, new sets of indicators will be put in place giving a more detailed insight into the underlying components of the current basic set. Users will have the possibility to make a print and to download the animation. Furthermore the look and feel will be improved, for instance by introducing a more visible time stamp during animation, a time slider instead of a time pull-down menu. Last but not least a BCC demo will be included for specialists, but also for more basic users in order to achieve a better understanding of economics in general and the business cycle in particular.

4. Some more analytical examples

 

Figure 4 highlights the fact that the cyclical stage of Gross Domestic Product (GDP) is somewhere in-between the cyclical stages of its building blocks Consumption, Investment, Imports and Exports. In an open economy such as the Netherlands, exports are often a driving force behind a recovery. Like in many other countries Dutch private consumption is lagging a bit behind, moderating GDP increases during recovery, but on the other hand moderating decreases (after GDP peak) as well.

 

Figure 4. Dutch GDP and its components          Figure 5. Dutch& French leading indicators

 

Figure 5 shows that for both France (red symbols) and the Netherlands (blue symbols) the typical behaviour of two indicators – Economic Sentiment Indicator and Industrial production – which are generally considered as having a lead in time as compared to GDP. In real time these two monthly indicators are also available much sooner as quarterly GDP, the latter being the end result of an in-depth statistical analysis and intensive cross-checking within National Accounting.

An animated comparison between EU25 and the US (see below) reveals that at the dawn of the new millennium (see figure 6.a) both were in the well performing "green" phase of the business cycle. At the end of 2000 the US indicators were over their peak, still above the long term trend, however, moving downwards, while EU25 was peaking (see figure 6.b).

 

Figure 6.a. EU25 and the US in Jan 2000          Figure 6.b. EU25 and the US in Dec 2000

 

Figure 6.c. EU25 and the US in Dec 2001         Figure 6.d. EU25 and the US in Jun 2005

 

One year later, end of 2001, EU25 was in a similar phase of the business cycle as the US was one year before. Meanwhile the US was heading further into a trough (see figure 6.c). It was not until mid 2005 that EU 25 was passing a trough, the US already being back on track heading towards a new peak (see figure 6.d) which was reached at the end of 2006.


5. Some methodological and technical notes  

 

5.1 Method of calculation

 

The graphical data is based on a re-calculation of seasonally adjusted statistical time series. Except for prices which are as usual practice not seasonally adjusted. All the series are either expressed as an absolute index or as an absolute number, i.e. not as a growth rate. As a first step in the calculation each of the series is de-trended. As a second step the cyclical component is extracted[2] by the Christiano-Fitzgerald (CF) band-pass filter. In order to make the cyclical series mutually comparable in terms of magnitude, each of the series is normalized so that each has an average of zero and a unit standard deviation. With this last step of calculation all series fit within the same graphical frame, without affecting the direction of cyclical development of the indicator.

 

Because of these various steps of re-calculation, the resulting values are only valuable within the graphical framework of the BCC. These cyclical values do not have their own meaning like the original statistics do[3]. Real statistical data, i.e. the original counterpart of the graphical values, are displayed on the right of the graph. As such the main purpose of the BCC graph is to help users in a visual way, to create interest, and to offer a doorway to economic key statistics of Europe.

 

5.2 The BCC application

 

The BCC is built on open sources, client-based and runs entirely within the Internet browser with just a few calls to Eurostat's server. The graphical display is Flash based. For the data input holding the time series of the indicators, an XML format is used. The back-end of the BCC system is also compatible with SDMX[4].

 

Twice a day (11:00 am and pm) the BCC checks for new data and recalculates its internal cyclical values accordingly.

 

In the medium-term Eurostat intends to make the BCC application generally available to other (statistical) organisations within the framework of the EU-PL rules of the European Commission. To a large extent the BCC settings and language(s) can be modified by changing the configuration files. 

 

 


ANNEX: Information behind the table, graph and map buttons of the BCC

 

The information which is opened when the users clicks on one of the buttons on the right of the indicators is shown in the screenshots presented below. The following screenshots are taken from the GDP indicator. Figure 7 opens with the simple default table of GDP, which the user can download, print, save, etc, or by simple clicking create graphs or maps (see figures 7.b. and 7.c.). These self-created graphs and maps can be printed and saved or bookmarked for later instant re-use. Here the user can also find a more elaborate methodological description of the indicator.

 

Figure 7.a. Table view on GDP for the latest 12 periods for all countries.

 

 

Figure 7.b. Customizable graphs of GDP          Figure 7.b. Customizable maps of GDP

 

 

 



[1] Acknowledgements go to Statistics Netherlands for sharing their visual concept of tracing business cycles.

[2] The CF filter is used as a default, with 2-10 years as lower and upper boundaries. These boundaries tend to yield rather smooth cyclical patterns, and implicitly relatively stable long term trends. Alternative methods, such as the Hodrick-Prescott filter, the Henderson filter, the Phase Average Trend method and others have been/will be under investigation, and some of them are/will be built into the back-end of the BCC application (as plug-in) for analytical purposes and fine-tuning of BCC filter settings. A methodological explanation of these filter methods can be found in Eurostat's working paper "Statistical Methods for Potential Output Estimation and Cycle Extraction" (not yet available as link).

[3] In real time earlier cyclical outcomes, i.e. their magnitude and relative position compared to the long term trend may change as new data streams into the series. These revisions, however, are expected to be limited for the chosen CF filter settings. Furthermore the graphical display concentrates on the direction of cycle evolution, instead of the magnitude of single cycle "observations". Fine-tuning of the BCC settings (see footnote 2) will aim at further reduction of these revisions.

[4] SDMX is an XML type of standard for international exchange of statistics, supported by OECD, IMF, ECB, Eurostat and others.