Unintended side effects of financial market interventions on banks and firms

Abstract

The economy is a complex system because market participants do not act independently but adjust their behavior to other agents and to the outcome which emerges from their joint actions (Arthur, 2014). Dependencies among participants can impede policy makers capabilities to influence or steer the course of the economy. Kambhu et al. (2007) argue that to influence developments in financial markets, for instance to prevent crises from spreading, there are only “coarse or indirect options” available for policy makers. Similar to crises which propagate through a complex system, interventions might result in unintended side effects which can also disseminate through the system. Thus, in a complex system, unintended consequences of policy efforts may well be the rule. Policy makers try to ward off or mitigate negative consequences for the economy and society during periods of crisis. For instance, during the Covid crisis large scale support programs for firms in Western economies were set up to avoid bankruptcies. Similarly, during the sovereign debt crisis in the Eurozone, the European Central Bank (ECB) set up large scale asset purchase programs as well as additionally longer-term refinancing operations (LTRO) which provided immediate support to financial market participants’ liquidity positions and thereby prevented a melt-down of the financial system. During these periods, immediate and abundant liquidity supply is of utmost importance. Meanwhile, crisis measures, due to their massive scale and non-specific target group, may entail unknown or unintended side effects for instance on competition among market participants, firms’ investment behavior, or changes in lending strategies and risk taking behavior of banks. Likewise, new regulatory frameworks such as the introduction of new markets can have consequences previously not thought of. For policy makers it is important to know direct effects of policy interventions but also to be aware of the possibility and impact of indirect or unexpected side effects in order to evaluate measures taken and to learn for future design of regulation or intervention. This thesis sheds light on the unintended side effects that followed policy interventions such as the introduction of new markets by the regulator or unconventional monetary policy measures. More specifically, in Paper 1, together with my co-authors, I study how banks respond in terms of their lending as well as risk-taking behavior when the regulator allows a market for covered bonds. Covered bonds reduce refinancing costs of mortgage loans and therefore make mortgage lending more profitable. Surprisingly, we observe that banks exposed to the regulation do not increase mortgage lending. Instead, we find that covered bonds increase total balance sheet liquidity which enables banks to extend more risky and less liquid firm lending. In Paper 2 and 3 I assess the unintended side effects of the first large asset purchase program of the ECB - the Securities Market Program (SMP). I show that asset purchases can cause spillover effects on investments across firms in Paper 2. In line with previous findings on peer effects between firms (e.g. Bustamante and Fr´esard, 2021; Dougal et al., 2015), I observe that firms adapt their investment decisions to affected peer firms. With this finding, I contribute to the understanding of a slowdown in economic recovery after large asset purchases as pointed out by Acharya et al. (2019). In Paper 3, together with my coauthor, I provide an explanation for the phenomenon of a slowing down of business dynamics among very small firms which began in 2010 in Germany. We show that small and medium sized enterprises (SMEs) and their plants have lower market exit probabilities when they were exposed to asset purchases during the SMP. In Paper 4, together with my co-authors, I demonstrate that banks which operate in many different regulatory regimes, i.e. which have a geographically complex organizational structure, show higher default risks and are more likely to receive state aid. The results indicate that a lack of international coordination in financial regulation might result in the unintended side effect that internationally operating banks increase their risk-taking behavior. Interrelations and connections across market participants such as bank-firm links, supply chains, demand factors or peer behavior, form the economy into a complex system. This poses challenges to empirically assess unintended consequences of policies on banks and firms. The researcher is faced with a dilemma of more complex empirical modelling, which can take at least some parts of the relationships between market participants into account but which is difficult to comprehend, versus simple but very reductive models which might not be able to capture interconnections because they rely on assumptions such as independently drawn observations or isolated treatment and control groups. In this thesis, I accommodate these challenges by choosing empirical strategies which are very much related to a common framework, which is difference-in-differences analysis, but extend it to allow for a more comprehensive understanding. The common difference-in-differences model is attractive and very popular due to its relative simplicity. The researcher compares a treatment group e.g. affected by a policy change, to a control group over time. However, there are strong assumptions underlying the difference-in-differences approach, for instance there must not be spillovers from one group to the other. To ease this assumption, in Paper 2, I extend the empirical model similar to Berg et al. (2021) and allow for spillover effects across firms which operate in the same region and industry. The extended version is comparable to previous difference-in-differences approaches but allows for somewhat more complex modelling to gain insights into potential biases due to spillovers. In Paper 1, we also adapt the common difference-in-differences framework to allow for time-varying differential effects to assess whether differential effects decay over time. This approach is in particular suitable for our empirical setting as we conjecture in this analysis that differences between treated and control group vanish over time. In Paper 4 we make the complexity of organizations the topic of research itself and can see that geographical complexity can lead to higher default risks. In this thesis I emphasize the causal identification of effects of policy shocks on banks and firms. This approach might suffer from taking little account of external validity. The price of a stringent causal analysis can be that the finding only holds for a sub-sample of firms. In Paper 1, we focus on the Norwegian economy, which might be a special case with its prolonged house price growth and its dependency on the oil market, among others. We try to provide generalizable arguments by adding a theoretical model from which we derive predictions. For instance, we learn that banks extend firm lending only if firm risk is sufficiently low. It might well be that in other countries firm risk is higher, and therefore the impact of the covered bond market on bank lending is different. In Paper 2, I restrict the sample of firms to SMEs which only have one bank. On the one hand, this allows me to draw conclusions on the group of firms which are highly innovative and important for the German economy - SMEs. On the other hand, it limits the informative value when judging on the whole economy including also larger firms. However, the results on SMEs’ behavior might be generalizable to other Western countries as long as the context in which firms operate is comparable. In Paper 4 we include almost all large European banks in the analysis which has the advantage that results apply to a wider setting. Nevertheless, in this set up we do not claim to find causal effects and restrict ourselves to a descriptive analysis. Conclusions drawn from assessing side effects should also take into account the intended effects of policy measures and whether these succeeded. Concerning the introduction of covered bond markets in Norway, the intention was to create a market for safe assets, i.e. assets which are low in risk and money-like. As a consequence, balance sheet liquidity of banks increased and therefore liquidity risks were lowered. The side effect that banks extend lending to firms while still becoming more stable institutions seems to be a positive effect to the Norwegian economy. Concerning the main and side effects of asset purchases during the sovereign debt crisis, we must note that for instance the SMP was very successful in achieving its main goal of lowering government bond yields (e.g. Gibson et al., 2016; Eser and Schwaab, 2016; Ghysels et al., 2016) and therefore in preventing a collapse of the Eurozone. Detrimental side effects as this thesis finds, have to be weighed against the success of the program. Policy makers can learn from this thesis the nature of side effects, such that they can decide whether they want to accept these, pursue countervailing measures, or take them into account when considering to set up similar programs at other times.