Daniel Porath

Does Diversification Improve the Performance of German Banks? Evidence from Individual Bank Loan Portfolios

Should banks be diversified or focused? Does diversification indeed lead to increased performance and therefore greater safety on the part of banks as traditional portfolio and banking theory would suggest? Recently, Acharya et al. (J Bus, 79:1355–1412, 2006) have found that for Italian banks the answers to these questions depend on the level of risk that a bank has taken. In this paper we investigate whether this result is robust to the choice of the sample and to the calculation of the risk variable. To this end we use a unique data set of the individual bank loan portfolios of 983 German banks for the period from 1996 to 2002 and calculate a Value-at-Risk based risk variable. We then investigate the link between banks’ profitability and their portfolio diversification across different industries, broader economic sectors and geographical regions. We find little evidence of large performance benefits associated with diversification: For the majority of our data, diversification tends to be associated with reductions in bank returns, even after controlling for risk. Only in a few cases (e.g., high-risk banks and industrial diversification) do we find statistically significant positive relationships between diversification and bank returns. Our findings contradict both the empirical findings of Acharya et al. (J Bus, 79:1355–1412, 2006) and the theoretical findings of Winton (Don’t put all your eggs in one basket? Diversification and specialization in lending. Working Paper No. 00-16, University of Minnesota, 1999).

Does Capital Regulation Matter for Bank Behavior? Evidence for German Savings Banks

The aim of this paper is to assess how German savings banks adjust capital and risk under capital regulation. We estimate a modified version of the model developed by Shrieves and Dahl (1992). In comparison to former research, we impose fewer restrictions with regard to the impact of regulation on capital and risk adjustments. Besides, we complement our analysis with dynamic panel data techniques and a rolling window approach. We find evidence that the coordination of capital and risk adjustments depends on the amount of capital the bank holds in excess of the regulatory minimum (the so-called capital buffer). Banks with low capital buffers try to rebuild an appropriate capital buffer by raising capital and simultaneously lowering risk. In contrast, banks with high capital buffers try to maintain their capital buffer by increasing risk when capital increases.

Estimating Probabilities of Default for German Savings Banks and Credit Cooperatives

Savings banks and cooperative banks are important players in the German financial market. However, we know very little about their default risk, because these banks usually resolve financial distress within their own organizations, which means that outsiders can-not observe defaults. In this paper I use a new dataset that contains information about financial distress and financial strength of all German savings banks and cooperative banks. The Deutsche Bundesbank has gathered the data for microprudential supervision. Thus, the data have never before been exploited for statistical risk assessment. I use the data to identify the main drivers of savings banks’ and cooperative banks’ risk and to detect structural differences between the two groups. To do so, I estimate a default pre-diction model. I also analyze the impact of macroeconomic information for forecasting banks’ defaults. Recent findings for the U.S. have cast some doubt on the usefulness of macroeconomic information for banks’ risk assessment. Contrary to recent literature, I find that macroeconomic information significantly improves default forecasts.

Statistical Methods to Develop Rating Models

The Internal Rating Based Approach (IRBA) of the New Basel Capital Accord allows banks to use their own rating models for the estimation of probabilities of default (PD) as long as the systems meet specified minimum requirements. Statistical theory offers a variety of methods for building and estimation rating models. This chapter gives an overview of these methods. The overview is focused on statistical methods and includes parametric models like linear regression analysis, discriminant analysis, binary response analysis, time-discrete panel methods, hazard models and nonparametric models like neural networks and decision trees. We also highlight the benefits and the drawbacks of the various approaches. We conclude by interpreting the models in light of the minimum requirements of the IRBA.

Scoring Models for Retail Exposures

Rating models for retail portfolios deserve a more detailed examination because they differ from other bank portfolios. The differences can mainly be attributed to the specific data structure encountered when analyzing retail exposures. One implication is that different statistical tools have to be used when creating the model. Most of these statistical tools do not belong to the banker’s standard toolbox. At the same time – and strictly speaking for the same reason – the banks’ risk management standards for retail exposures are not comparable to those of other portfolios.

Applying the Bass model to pharmaceuticals in emerging markets

Albeit the Bass model was not designed for predicting sales of newly launched drugs, pharmaceutical companies commonly use it for this purpose, mainly because of its good predictive power. Empirical experience, however, mainly refers to mature markets and it is unclear how the model behaves in emerging markets. We try to fill this gap in the literature by comparing the estimation results of the Bass model between emerging markets and mature markets in a big dataset including more than 5,000 new launches from different countries. Our results show a good performance of the model in emerging markets. Compared to mature markets the estimated parameters on average are the same, but there is a higher heterogeneity between individual countries. Our findings favour the application of the Bass model in emerging markets, but also highlight the importance of selecting individual parameters for each country and therapeutic class.

Size and dynamics of order-of-entry effects in pharmaceutical markets:

In anticipation of long-run competitive advantages, pharmaceutical companies often try to enter new markets earlier than their competitors. It is questionable, however, whether this effort is really justified, because knowledge about the expected size of these order-of-entry advantages is scarce. Apart from confirming the existence of long-run advantages for the first mover for some selected brands, the literature gives hardly any guidelines. The objective of this article is to provide the information on the existence, the size, and the dynamics of order-of-entry effects, which is necessary for deciding on the optimal moment for a launch. The author proposes a new framework—a fractional panel probit model—which is fitted to a big and representative sample. With this procedure, order-of-entry effects can be measured in a multivariate model and as absolute market share differences, which is more informative for pharmaceutical companies than former measures.