Nowcasting Germany’s stagnant industrial production as the nation debates its economic model

Being Europe’s largest economy has always come with a lot of scrutiny. Even so, the recent deterioration of German macro data has received more attention than usual, reigniting political debates about inflation, climate change and the future of Europe’s industrial sector. 

The German economic model of the last 30 years has been heavily criticised in the media during the past two years. The discussion has centred on Germany’s reliance on cheap energy from Russia, low domestic wage growth and foreign export markets to sustain somewhat reasonable economic growth. While the energy crisis has been global, nowhere has it opened up so many questions about the status quo as it has in Germany. 

Additionally, when it comes to the impact of supply-chain disruptions on the economy, Germany has suffered the most within the eurozone. In the first quarter of 2022, a record-breaking 90 percent of German companies surveyed by the European Commission saw equipment shortages limiting their production, compared to 54 percent for the eurozone in general. And while the easing of supply-chain bottlenecks since then has been welcome news, a combination of falling new orders and a weaker global consumer have resulted in a stagnant German economy.

Elevated policy uncertainty, significant data deviations from consensus forecasts and a potential pandemic-induced macro regime shift in Germany have made tracking – and forecasting – economic data more important.

Selecting the data to build a nowcast

A combination of leading (soft) and lagging (hard) data can help us gauge where the economy is headed in the short and medium term. To get a first look at how Germany is doing on a monthly basis, we can create an unweighted index by standardising industrial production, new orders, retail sales and exports. 

This hard data proxy (shown in the first chart, and which can be replicated for other countries) has been tracking GDP growth quite well (as the second chart shows), demonstrating how these variables have developed on a relative basis to each other at a glance. 

Going a step further in our analysis by incorporating survey-based indicators can give us a better understanding of how these lagging data points might change in the near future. 

We incorporated a mixture of three survey-based indicators into our forecasts: the business, consumer, and economic expectations indices from the Ifo, GfK and ZEW institutes. All three highlight how German companies and consumers are evaluating their current situation and what they are expecting for the future. 

In addition to looking at the headline indicators, most survey reports are accompanied by various sub-indicators, which are helpful in determining trends for certain industries. In the case of industrial production, simply focusing on the production expectations of German companies surveyed by the Ifo institute can be a helpful guide before one attempts to build a model. 

In our two charts below, we can see that a negative bias for the German economy remains in place. Production expectations continue to point to a weakening of industrial activity, while our soft data proxy, incorporating the above-mentioned surveys, has been in negative territory for 17 consecutive months. 

As most hard data points have stagnated and remained in negative territory, GDP growth has been lacklustre for the past three quarters. Forward-looking indicators appear to have weakened as well, following the recovery at the beginning of 2023. Both the “soft” and “hard” macro data point to a subdued outlook, and leave us with a slight negative bias for the German economy going into the second half of the year. 

While we are not expecting a severe contraction for GDP, signs of a strong recovery are broadly missing. 

More about the Convera nowcast model

Nowcasting became more important during Covid-19. With so much disruption in the short term, “predicting” the recent past became more attractive given the significant time lag before so many conventional economic indicators are published. 

At Convera, the Market Insights team used regression analysis to estimate a nowcast model for the German industrial production index using Indicio, a software tool that estimates univariate and multivariate time-series models to forecast time-series data. 

From a selection of 30 time series (a combination of leading, macro, and financial variables), four indicator variables were identified as having the highest influence coefficients using a coefficient search strategy: the unemployment rate, the New York Fed’s Global Supply Chain Pressure Index, terms of trade and the OECD’s leading economic indicator (LEI) for the manufacturing sector. Given the variable selection and subsequent data availability, the model uses data from January 2000 onwards. The model is supported by mixed frequency structures, which solve the problem of losing potentially useful timelier information. 

Exogenous variables have been included to account for sudden events, such as the global financial crisis (a demand-side shock) and the pandemic (a supply-side shock), which may have disrupted the long-term stable dynamics between variables in the short run but have not caused a structural break in the series. The final model is a weighted average of 23 univariate and multivariate models using RMSE stepwise criteria. 

The results show that the upcoming June print of German industrial production index is forecast to fall to 98.2, down from the current 98.6 level – which would bring the index to its lowest level since December 2022. The forecast implies a 0.4 percent month-on-month decline – which would be the second consecutive monthly drop since April - and a 0.1 percent year-on-year decline, the first on an annual basis in five months. 

Our forecast is more downbeat than the -0.3%* month-on-month consensus, but in line with the negative trend expected by the market.

The SHapley Additive exPlanations (SHAP) method is used to disaggregate the contribution of each variable to the final forecast. The continuing deterioration in in the OECD’s LEI for the manufacturing sector, from 100.24 to 99.85 (the lowest level since August 2020), had the largest single downward contribution. 

A minor uptick in the Global Supply Chain Pressure Index, rising from -1.56 in May to -1.2 in June, also had a significant downward effect. While unemployment increased from 2.9 percent in May to 3.0 percent in June, the effects of this variable are felt with a 5- to 8-month lag, and thus a falling unemployment rate during the December 2022 – February 2023 period had a positive upward effect on the forecast. 

Finally, terms of trade, while showing an improvement from 96.8 in May to 98.1 in June, had a negligible impact.