What is KOF Economic Barometer?
The KOF Economic Barometer, from the KOF Swiss Economic Institute at ETH Zurich, tracks the Swiss business cycle. It runs monthly since 1976 and tracks the Swiss real GDP growth rate cycle in Version 2014.
It blends over 500 series, mostly business surveys, plus customs, official and foreign data. It is not a GDP forecast, price or jobs gauge, nor the global gauge or Business Situation Indicator. Arabic media use KOF Economic Indicator and Swiss business confidence, while 100 marks the 10 year average, with higher values favourable and lower values subdued.
When is it released and what happens each release?
KOF calculates the Barometer in the last week of the reference month, shifting late input series forward to complete the panel. It publishes on a last working day at 9:00 a.m. Zurich time.
The yearly vintage normally arrives with the September release but moves to October in exceptional years. Each release brings an English and German press release, chart, 12 month table and data files. History changes only with input revisions, while each yearly vintage revises all history.
What moves KOF Economic Barometer?
No fixed weights exist; contributions are grouped into bundles, with manufacturing usually the largest driver.
- Manufacturing: captures orders, production, employment, business situation and stocks of intermediate goods across sub industries.
- Construction: captures building activity, demand expectations, order backlog, capacity use and hiring plans.
- Financial and insurance services: covers banks and insurers, with business situation, demand, income, lending and claims.
- Other services and hospitality: covers transport, information, communication, business and personal services plus hospitality, tracking demand, capacity, prices and obstacles.
- Foreign demand: captures export orders and surveys from trading partners.
- Private consumption: captures retail, new vehicle registrations and consumer surveys.
In March 2020 it fell 10.1 points to 91.7. On April 30, 2020 it fell 28.2 points to 63.5, its largest drop and near the 2009 crisis level. April saw nearly all groups down, led by manufacturing and other services.
On May 29, 2020 it fell 6.5 points to 53.2, its historic low and about half the February 101.8. Manufacturing and foreign demand stayed clearly negative in May, while private consumption, construction and order backlogs sent slightly improved signals.
How is KOF Economic Barometer calculated?
- Reference series: Take SECO, meaning the State Secretariat for Economic Affairs, seasonally adjusted real quarterly GDP. Interpolate it to monthly levels with the Denton method, then smooth growth with a 13 month centred average.
- Pool: Preselect around 500 plausible series and fix the expected correlation sign. Allow levels, differences and balance shares, assign quarterly series to all 3 months and average daily series monthly, avoiding double counting by industry code.
- Seasonal adjustment: Run each series and transformation through X-12-ARIMA. Seasonal factors stay frozen for the vintage.
- Publication lag alignment: Shift late series forward by vertical realignment and freeze lags until next vintage. Fill gaps from dead series with EM imputation capped to avoid extremes.
- Selection: Over a 10 year window, check correlations up to 7 months each way and require full coverage. Keep only series with the expected sign, a 0 to 6 month lead and 5% significance. Pick the best form with a utility score rewarding longer leads with diminishing returns.
- Aggregation: Extract the first principal component from scaled series through the correlation matrix, with no final smoothing. For example, loadings of 0.60, 0.55 and 0.45 on values of 0.8, minus 0.2 and 0.4 give 0.55. With a spread of 1.10, this scales to 0.50 and a Barometer of 105.0.
- Scaling: Standardise the factor to SD 1, setting Barometer to 100 plus 10 times the factor. This gives mean 100 and SD 10 in sample.
- Out of sample use: Apply frozen seasonal factors, in sample means, spreads and loadings to new data. Extend early history with the EM method holding loadings fixed.
Key point: 100 marks average pace, not zero growth, not a PMI 50 boom bust line and not a GDP forecast. It signals the stance of the cycle as early as possible and often turns before yearly GDP.
