ifo Economic Forecast

Economic forecasts are estimates of future economic activity in the overall economy, with a focus on the slowing or speeding up as well as the turning points of economic variables in the course of the business cycle. Economic fluctuations are expressed in terms of the development in real quarterly GDP.

The ifo Institute forecasts real and nominal GDP in Germany in terms of the demand components, because in matters of economic activity it is not an issue of trend growth, which requires a supply-side explanation, but of the variations around the trend. The forecast distinguishes between the demand of private households, the state, investors (equipment and buildings) and other countries (exports) and takes into account that these demand components are also affected in part by imports. Taking into account the sectoral business surveys (ifo KT) the forecast is broken down to the major sectors of the economy (IFOCAST). Every GDP forecast is supplemented by an estimate of primary distribution (employee compensation, corporate and investment income) as well as income redistribution by the state. In addition the government account (revenue, expenditure and net lending/net borrowing) is quantified and the foreseeable development on the labour market (number of employees, number of unemployed) is forecast in conjunction with the expected state of economic activity.

Partial forecasts are presented by using the system of national accounts and converted to half-year or yearly values for publication. In a multi-phased, iterative process the partial forecasts of GDP, the labour market and government accounts are tested for their economic plausibility and synchronized with each other until a computationally coherent forecast with the greatest subjective probability is achieved.

The ifo Institute employs a number of modern methods for forecasting economic activity such as vector autoregressive approaches (VAR models), econometric structural models and trend-cycle decomposition. This ensures that all information efficiently flows into the forecast and that the expert knowledge of the forecaster is adequately incorporated. As a long-standing practice, the ifo Institute regularly evaluates the quality of its own forecasts of economic activity. It comments on important conceptual changes in the system of national accounts, in which accounting system the Ifo economic forecasts are embedded.

Accurate economic forecasts are desirable, but challenging. Like all forecasts, the ifo Institute's economic forecasts deviate to a certain extent from the official numbers published later. In order to ensure maximum transparency, the ifo Institute's economic forecasts, which have been produced in June and December since 1991, are documented and evaluated on an ongoing basis. The following analysis focuses on the forecast of the annual rate of change of the price-adjusted gross domestic product (GDP) for the current and the coming year.

Since the length of the forecast horizon has a decisive influence on the accuracy of the forecast, in the following the forecasts are considered separately according to the respective forecast horizon. In order to obtain the GDP growth rate for the current year in December, only GDP in the fourth quarter needs to be forecast, since three quarterly GDP values in the current year are already known at this point. For the forecast of the GDP growth rate in the coming year, five quarters have to be forecast at this point: the GDP in the fourth quarter of the current year as well as the four quarters of the coming year. In the June forecast for the current year, three quarters of the current year still need to be forecast, and to obtain the value for the coming year, the forecast horizon is seven quarters.

The following figure shows the forecast error for different forecast horizons in quarters, calculated as the forecast value of the annual rate of change in GDP for a year minus the value for this rate published by the Federal Statistical Office at the beginning of the following year. If the forecast error is positive, the ifo forecasts were too optimistic; if it is negative, they were too pessimistic. The largest errors always were made in forecasts with a longer horizon, when singular and crisis-like events occurred. These include German reunification in the early 1990s, the bursting of the New Economy bubble in the early 2000s, the world financial and euro crises in the years after 2008, and the Covid and energy crisis in the early 2020s.

The average forecast error shows whether the forecaster makes systematic errors. Ideally, the average forecast error should be zero. In such a case, the errors occur purely at random and balance out on average over a longer period of time: the forecasts are unbiased and the forecaster is neither too optimistic nor too pessimistic. Planning based on such unbiased forecasts is thus always correct on average over many years.

The following figure shows that the ifo economic forecasts are unbiased at short forecast horizons. While the average forecast error is actually zero for a forecast horizon of one quarter, it is -0.04 percentage points for a horizon of three quarters. Statistically, however, this value is not significantly different from zero, as it lies within the 95% confidence interval. As the forecast horizon increases, the ifo business cycle forecasts tend to become too optimistic. This reflects in particular the mostly abrupt economic contractions associated with the crises, which cannot be anticipated a year in advance or even earlier.

In addition to unbiasedness, accuracy is another important criterion in the evaluation of forecasts. Since positive and negative deviations may cancel each other out when calculating the average forecast error, the mean absolute forecast error (MAE) is used to determine the forecast accuracy. Here, the forecast errors are included without a sign, so that the value can be interpreted as the average deviation of the forecast from the actual value (whether upwards or downwards) and thus as a measure of the accuracy.

Obviously, the forecast accuracy depends on the level of information at the time of forecasting. The following figure shows that the MAE is 0.07 percentage points if only one quarter has to be forecast. Thus, the ifo economic forecasts in December deviate on average by 0.07 percentage points from the annual GDP growth rate published by the Federal Statistical Office in the following January. As the number of quarters to be forecast increases, so does the uncertainty. With seven quarters to forecast, the MAE of the ifo economic forecast is 1.70 percentage points.

In order to form a final judgement on the value of the ifo economic forecasts, their accuracy must be compared with alternative forecasts. To determine the contribution of the knowledge of the ifo experts, it is useful to compare them with a naive forecaster who has no special knowledge of economic analysis or forecasting and who makes his forecasts with little effort. At the time of the forecast, the naive forecaster calculates the average of the quarterly growth rates of the price-, seasonally- and calendar-adjusted gross domestic product for the past ten quarters and simply extrapolates this from the current quarter to the end of the forecast period. Across all forecast horizons, the MAE of the ifo economic forecasts is lower than that of the naive forecaster. With only one quarter to forecast, the MAE is reduced by 61 percent, and with seven quarters to forecast, it is reduced by 12 percent. Of course, the forecasts of other experts could also be used as an alternative to the ifo economic forecasts. Therefore, the ifo economic forecasts are compared in the following with the so-called consensus forecasts, which were published at the same time as the ifo economic forecasts. Since 1989, Consensus Economics regularly asks about 30 experts from banks, investment trusts and economic research institutes (including the ifo Institute) about their forecast for the annual rate of change of important macroeconomic variables, such as price-adjusted GDP. The consensus forecast is usually understood as the average of the forecast values given by the individual experts. For shorter forecast horizons, the ifo economic experts significantly outperform the consensus forecasts. For example, the MAE of the ifo December forecasts for the current year is only about half as large as that of the consensus forecasts. With three and five quarters to be forecast, the improvement in the MAE is still 17 percent and 8 percent respectively. Only with a forecast horizon of seven quarters are the consensus forecasts superior. Overall, it can be concluded that the ifo economic forecasts make a measurable contribution to reducing uncertainty about the future development of the economy.

The Main National Accounts Data (Archive since 1974)  (in German)

• Tabellenanhang der ifo Konjunkturprognosen 1974-1979 (PDF)
• Tabellenanhang der ifo Konjunkturprognosen 1980-1989 (PDF)
• Tabellenanhang der ifo Konjunkturprognosen 1990-1999 (PDF)
• Tabellenanhang der ifo Konjunkturprognosen 2000-2009 (PDF)
• Tabellenanhang der ifo Konjunkturprognosen 2010-2019 (PDF)
• Tabellenanhang der ifo Konjunkturprognosen 2020-2026 (PDF)

The ifo Economic Forecast Data is also accessible via the EBDC website.

Quarterly Components of Real Gross Domestic Product by Expenditure – Archive since Winter 2007 (in German)

• Quarterly Forecasts Archive Summer 2026 (xls)

The ifo Economic Forecast Data is also accessible via the EBDC website.