The paper analyzes the forecasting performance of leading indicators for industrial production in Germany. We focus on single and pooled leading indicator models both before and during the financial crisis. Pairwise and joint significant tests are used to evaluate single indicator models as well as forecast combination methods. In addition, we investigate the stability of forecasting models during the most recent financial crisis.

This paper analyses the recession in 2008/2009 in Germany. This recession is very different from previous recessions in particular regarding their causes and magnitude. We show to what extent forecasters and forecasts based on leading indicators fail to detect the timing and the magnitude of the recession. This study shows that large forecast errors for both expert forecasts and forecasts based on leading indicators resulted during this recession which implies that the recession was very difficult to forecast. However, some leading indicators (survey data, risk spreads, stock prices) have indicated an economic downturn and hence, beat univariate time series models. Although the combination of individual forecasts provides an improvement compared to the benchmark model, the combined forecasts are worse than several individual models. A comparison of expert forecasts withthe best forecasts based on leading indicators shows only minor deviations. Overall, the range for an improvement of expert forecasts in the crisis compared to indicator forecasts is small.

We develop an evolutionary algorithm to estimate Threshold Vector Error Correction models (TVECM) with more than two cointegrated variables. Since disregarding a threshold in cointegration models renders standard approaches to the estimation of the cointegration vectors inefficient, TVECM necessitate a simultaneous estimation of the cointegration vector(s) and the threshold. As far as two cointegrated variables are considered, this is commonly achieved by a grid search. However, grid search quickly becomes computationally unfeasible if more than two variables are cointegrated. Therefore, the likelihood function has to be maximized using heuristic approaches. Depending on the precise problem structure the evolutionary approach developed in the present paper for this purpose saves 90 to 99 per cent of the computation time of a grid search.

The present paper compares expected inflation to (econometric) inflation forecasts based on a number of forecasting techniques from the literature using a panel of ten industrialized countries during the period of 1988 to 2007. To capture expected inflation, we develop a recursive filtering algorithm which extracts unexpected inflation from real interest rate data, even in the presence of diverse risks and a potential Mundell-Tobin-effect.

The extracted unexpected inflation is compared to the forecasting errors of ten

econometric forecasts. Beside the standard AR(p) and ARMA(1,1) models, which

are known to perform best on average, we also employ several Phillips curve based approaches, VAR, dynamic factor models and two simple model avering approaches.

Combining forecasts, we analyse the role of information flow in computing short-term forecasts up to one quarter ahead for the euro area GDP and its main components. A dataset of 114 monthly indicators is set up and simple bridge equations are estimated. The individual forecasts are then pooled, using different weighting schemes. To take into consideration the release calendar of each indicator, six forecasts are compiled successively during the quarter. We found that the sequencing of information determines the weight allocated to each block of indicators, especially when the first month of hard data becomes available. This conclusion extends the findings of the recent literature. Moreover, when combining forecasts, two weighting schemes are found to outperform the equal weighting scheme in almost all cases. Compared to an AR forecast, these improve by more than 40% the forecast performance for GDP in the current and next quarter.

We study how idiosyncratic income risk shapes the aggregate and distributional effects of labor and capital income taxation in dynamic general equilibrium models. To this end, we compare a heterogeneous-agent (HA) model with uninsurable idiosyncratic labor productivity risk and a ten-representative-agent (TE) model in which households correspond to fixed wealth deciles without such risk. At the aggregate level, both models generate qualitatively similar responses; however, the HA model exhibits a smaller recessionary impact driven by precautionary savings behavior, which stabilizes investment. At the distributional level, the models differ sharply. In the HA framework, tax shocks trigger endogenous mobility across wealth deciles. These inter-decile transition dynamics tend to benefit lower deciles. In contrast, the TA model features fixed household positions. Our findings highlight that while simpler multi-representative-agent models can approximate aggregate dynamics well, they may miss important distributional adjustment channels. The relevance of these mechanisms ultimately depends on the empirical importance of mobility across the wealth distribution, pointing to a key trade-off between model simplicity and accuracy.

Many countries and regions remain below the level of economic activity of the world’s most advanced economies. Some countries form growth clubs, some are stuck in the middle-income trap, and some stay on a very low level of economic activity. Although this situation is well documented on the country level, there is less evidence at the sub-national level within countries. We estimate county-level capital stocks and price indices and provide a comprehensive county-level data set for Germany. We find no evidence of convergence across all counties even if we condition on important drivers of long-term growth such as physical and human capital accumulation. Instead, we identify five convergence clubs, using endogenous clustering. We analyze differences in growth paths and describe the identified clusters based on variations in contributions of capital, labor, and total factor productivity to economic growth. Additionally, we examine the role of migration for regional development and find that net migration has in particular contributed to growth in richer regions.

Forecasts that minimize mean squared forecast error (MSE) often exhibit excessive volatility, limiting their practical applicability. We address this accuracy-smoothness trade-off by introducing a Multivariate Smooth Sign Accuracy (M-SSA) framework, which extracts smoothed components from leading indicators to enhance the signal-to-noise ratio and control the forecast volatility and timing. Applied to quarterly German GDP growth, our method yields smoothed forecasts that can improve forecasting accuracy, particularly over medium-term horizons. We find that while smoother forecasts tend to lag slightly around turning points, this can be offset by adjusting the forecast horizon. These findings highlight the practicality of the M-SSA framework for both forecasters and policymakers, especially in settings where forecast revisions or policy adjustments are costly.

Accurate macroeconomic forecasts are essential for effective policy decisions, yet their precision depends on the accuracy of the underlying assumptions. This paper examines the extent to which assumption errors affect forecast accuracy, introducing the average squared assumption error (ASAE) as a valid instrument to address endogeneity. Using double/debiased machine learning (DML) techniques and partial linear instrumental variable (PLIV) models, we analyze GDP growth forecasts for Germany, conditioning on key exogenous variables such as oil price, exchange rate, and world trade. We find that traditional ordinary least squares (OLS) techniques systematically underestimate the influence of assumption errors, particularly with respect to world trade, while DML effectively mitigates endogeneity, reduces multicollinearity, and captures nonlinearities in the data. However, the effect of oil price assumption errors on GDP forecast errors remains ambiguous. These results underscore the importance of advanced econometric tools to improve the evaluation of macroeconomic forecasts.

We show that the response of banks’ net interest margin (NIM) to monetary policy shocks is state dependent. Following a period of low (high) Federal Funds rates, a contractionary monetary policy shock leads to an increase (decrease) in NIM. Aggregate economic activity exhibits a similar state-dependent pattern. To explain these dynamics, we develop a banking model in which social interactions influence households’ attentiveness to deposit interest rates. We embed that framework within a nonlinear heterogeneous-agent NK model. The estimated model accounts well quantitatively for our key empirical findings.