Reinforcement learning policy recommendation for interbank network stability

This paper aims to shed light on the emergence of systemic risk in credit systems. By developing an interbank market with heterogeneous financial institutions granting loans on different network structures, we investigate which market architecture is more resilient to liquidity shocks and how the risk spreads over the modeled system. In our model, credit linkages evolve endogenously via a fitness measure based on different banks' strategies. Each financial institution, in fact, applies a strategy based on a low interest rate, a high supply of liquidity or a combination of them. Interestingly, the choice of the strategy influences both the banks' performance and the network topology. In this way, we are able to identify the most effective tactics adapt to contain contagion and the corresponding network topology. Our analysis shows that, when financial institutions combine the two strategies, the interbank network does not condense and this generates the most efficient scenario in case of shocks.

Could 'it' happen here?Australia survived the global financial crisis relatively unscathed, despite much higher interest rates than in other developed countries (see Figure 1). But our exceptional status may be short-lived. In May 2015, the cash rate was cut to 2 per cent, the lowest level since the current series began in 1990, and by some measures the lowest short-term rate since at least 1960. Nevertheless, unemployment has risen steadily, despite the fact that investment in the resources sector has yet to fall back to normal levels.It would seem prudent, then, to be prepared for interest rates to approach zero, as has already occurred in most other OECD countries.2 Sheehan and Gregory (2013) and Freebairn and Corden (2013) have called for increased infrastructure spending on this basis. Yet, as these authors acknowledge, such a policy faces considerable challenges. Furthermore, if warnings of 'secular stagnation' have any validity, near-zero interest rates may not be a once-in-a-generation emergency to be dealt with using ad hoc expedients, but an increasingly common situation requiring a more systematic response. In this case, changes to monetary policy would appear more desirable.This paper considers various means by which monetary policy may stimulate aggregate demand when its usual instrument - the short-term interest rate - is unavailable. The objectives of monetary policy are taken here to be the conventional ones of a stable value of money and full employment, analysed under the broadly 'Keynesian' assumption that fluctuations in nominal spending have significant effects on real output. Financial stability objectives are assumed to be dealt with separately via (macro)prudential regulation.3The paper begins by describing the problem of the zero lower bound, and the logic of expectations management in general and level targeting in particular as a solution. It then argues, however, that level targeting requires a more careful choice of target variable than the current regime of inflation targeting, and that the price level, nominal GDP and nominal wages all appear problematic. A more eclectic mix of policies is then considered.The zero lower bound, and the vital role of expectationsAs long as money can be stored at negligible cost, interest rates cannot fall below zero, because hoarding money would then be a superior alternative to lending it. This poses a problem for conventional monetary policy, under which the usual response to falling inflation or rising unemployment is a lower interest rate. What happens if the interest rate cannot be cut any further? The experience of Japan in the 1990s, and much of the world since 2008, shows that near-zero interest rates are perfectly compatible with low and falling inflation, high and rising unemployment, and output well below previous estimates of potential.Can monetary policy still be effective in such a situation? Theory suggests that expectations management is the key. In the standard permanent income or life-cycle model, current spending is determined by expected future income and interest rates. If promises about future policy are successful in changing these expectations, they may increase demand without any immediate change in the central bank's balance sheet. Conversely, even a very large balance sheet expansion may be ineffective if it is believed to be temporary and thus does not change expectations.4 Krugman (1998), motivated by the then unusual experience of Japan, showed that, while a temporary monetary expansion would fail to raise prices and output at the zero lower bound (since it cannot change the current interest rate, and, being temporary, will not affect any future variables), a credible permanent expansion could work, by increasing the expected future price level and therefore reducing the real interest rate. (This assumes that a permanently higher money supply must eventually create proportionally higher prices at some point in the future when the interest rate rises above zero. …

In order to realize the sustainable development of the communication industry, many banks have developed some new projects to provide loan to the mobile phone brand companies (MPBCs). This paper studies a perfect information game among three parties: telecom operator (TO), MPBC, and bank. In the first stage, the bank decides an interest rate and shows it to the TO and the competitive MPBC. In the second stage, the TO and MPBC are engaged in Cournot games: a simultaneous subgame, a TO-as-leader sequential subgame, and an MPBC-as-leader sequential subgame. The TO’s and the MPBC’s decisions and the production/sale quantities are investigated. The impacts of the interest rate and the substitute factor on the TO’s and the MPBC’s optimal decisions are analyzed. When the substitute factor is high, at a low interest rate, the total sales in the simultaneous subgame is higher than those in the other two subgames; at a high interest rate, the total sales in the MPBC-as-leader subgame is higher than those in the other two subgames. However, when the substitute factor is low, at a low enough interest rate, the total sales in the simultaneous subgame is higher than those in the other two subgames; at a high enough interest rate, the total sales in the MPBC-as-leader subgame is higher than those in the other two subgames; at a moderate interest rate, the total sales in the TO-as-leader subgame is higher than those in the other two subgames. Besides, the optimal interest rate of the bank is investigated and the impact of the substitute factor on the optimal interest rate is analyzed. The bank sets a higher interest rate in the MPBC-as-leader subgame than those in the other two subgames. Besides, when the substitute factor is low, the bank sets a lower interest rate in the TO-as-leader subgame than that in the simultaneous subgame; however, when the substitute factor is high, the bank sets a higher interest rate in the TO-as-leader subgame than that in the simultaneous subgame.

Interbank Networks in the Shadows of the Federal Reserve Act

We analyze how interest rates affect cross-border portfolio investments. Data on U.S. bond holdings by foreign investors from 31 countries for the period 2003 - 2016 and a large variety in movements in interest rates in these countries provide for a unique way to analyze shifts in investment behavior in response to interest rates. We find that low(er) interest rates, now prevailing in many advanced countries, lead to greater investment in general into the United States, with the effects generally driven by investment in (higher yielding) corporate bonds, rather than in Treasury bonds. In addition to affecting overall investments, lower interest rates at home are associated with a greater weight on corporate bonds, consistent with search- for-yield. The results are economically important and robust to controlling for a number of country-specific macroeconomic and financial conditions as well as to sample restrictions and choices of interest rate. Our findings have important policy implications in that they suggest that low interest rates can lead to shifts in the volume and composition of overseas investments.

Nominal interest rates may remain substantially below the averages of the last half century, because central banks' inflation objectives lie below the average level of inflation, and estimates of the real interest rate that are likely to prevail over the long run fall notably short of the average real interest rate experienced during this period. Persistently low nominal interest rates may lead to more frequent and costly episodes at the effective lower bound (ELB) on nominal interest rates. We revisit the frequency and potential costs of such episodes in a world of low interest rates, using both a dynamic stochastic general equilibrium (DSGE) model and the Federal Reserve's large-scale econometric model, the FRB/US model. Four main conclusions emerge. First, monetary policy strategies based on traditional, simple policy rules lead to poor economic performance when the equilibrium interest rate is low, with economic activity and inflation more volatile and systematically falling short of desirable levels. Moreover, the frequency and length of ELB episodes under such policy approaches are estimated to be significantly higher than in previous studies. Second, a risk adjustment to a simple rule—whereby monetary policymakers are more accommodative, on average, than prescribed by the rule—ensures that inflation averages its 2 percent objective, and requires that policymakers systematically seek inflation near 3 percent when the ELB is not binding. Third, commitment strategies, whereby monetary accommodation is not removed until either inflation or economic activity overshoots its long-run objective, are very effective in both the DSGE and FRB/US models. And fourth, our simulation results suggest that the adverse effects on economic and price stability associated with the ELB may be substantial at inflation targets near 2 percent if the equilibrium real interest rate is low and monetary policy follows a traditional approach. Whether such adverse effects could justify a higher inflation target depends upon the degree to which monetary policy strategies that differ substantially from such traditional approaches are feasible, and an assessment of a broader array of the inflation target's effects on economic welfare.

Using a unique dataset of the Euro-area and the U.S. bank lending standards, we find that low (monetary policy) short-term interest rates soften standards for household and corporate loans. This softening--especially for mortgages--is amplified by securitization activity, weak supervision for bank capital, and low monetary policy rates for an extended period. Conversely, low long-term interest rates do not soften lending standards. Finally, countries with softer lending standards before the crisis related to negative Taylor rule residuals experienced a worse economic performance afterward. These results help shed light on the origins of the crisis and have important policy implications. The Author 2011. Published by Oxford University Press on behalf of The Society for Financial Studies. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com., Oxford University Press.

Reinforcement learning (RL) is one of the popular methods for intelligent control and decision making in the field of robotics recently. The goal of RL is to learn an optimal policy of the agent by interacting with the environment via trail and error. There are two main algorithms for RL problems, including model-free and model-based methods. Model-free RL is driven by historical trajectories and empirical data of the agent to optimize the policy, which needs to take actions in the environment to collect the trajectory data and may cause the damage of the robot during training in the real environment. The main different between model-based and model-free RL is that a model of the transition probability in the interaction environment is employed. Thus the agent can search the optimal policy through internal simulation. However, the model of the transition probability is usually estimated from historical data in a single environment with statistical errors. Therefore, an issue is faced by the agent is that the optimal policy is sensitive to perturbations in the model of the environment which can lead to serious degradation in performance. Robust RL aims to learn a robust optimal policy that accounts for model uncertainty of the transition probability to systematically mitigate the sensitivity of the optimal policy in perturbed environments. In this overview, we begin with an introduction to the algorithms in RL, then focus on the model uncertainty of the transition probability in robust RL. In parallel, we highlight the current research and challenges of robust RL for robot control. To conclude, we describe some research areas in robust RL and look ahead to the future work about robot control in complex environments.

We study the relation between IPO investment and the rate of interest. We model the IPO timing decision and show that the implied relation between interest rates and investment is non-monotonic, and the data support the implication. At low rates of interest firms delay their IPOs. This happens because during the pre-IPO period the firm forgoes earnings that do not matter as much at low interest rates. The 1950's and early 1960's, especially, were periods of very low real interest rates, and IPO investment was low, with firms delaying their IPOs significantly. A qualitative difference seems to exist between investment of IPO-ing firms and the investment of incumbent firms which is decreasing in the interest rate, as neoclassical theory predicts.

In recent years, the importance of international capital flows has been increasing, which has significantly accelerated the integration process of the international financial system. After 2008, the central banks of major western developed economies implemented loose monetary policy, which led to a large amount of capital flowing into new market countries. However, international capital flowed out of new market countries after Federal Reserve Chairman Ben Bernanke said in 2013 that the Federal Reserve would gradually withdraw from quantitative easing.The great fluctuation of international capital has aroused the debate about monetary policy and macro-prudential policy. Rey (2015) believes that under any exchange rate system, only the control of international capital flows can maintain the independence of monetary policy. In view of the above findings, since 2016, the People’s Bank of China has established a macro-prudential regulatory mechanism for international capital flows to resist external financial shocks.By introducing domestic and international financial institutions, this paper brings the macro-prudential policy of foreign financial shocks and capital flows into the DSGE framework of small countries’ open economy. Domestic financial institutions can raise funds by absorbing the deposits of their own residents and foreign residents; international financial institutions buy and sell the creditor’s rights issued by various countries, so as to implement capital flows.It is found that under the impact of foreign interest rate increase, on the one hand, it will tighten the balance sheet of domestic financial institutions, and reduce the scale of corporate credit through bank incentives and constraints; on the other hand, it will cause the devaluation of domestic currency and international capital outflows, and affect the stability of the financial system. The macro-prudential management mechanism of capital flows can enhance the ability to resist external financial shocks. For China, the proportion of foreign debt is relatively low, and the policy effect is relatively mediocre for the counter cyclical tax adjustment mechanism of external borrowing of domestic financial structure; however, the counter cyclical adjustment mechanism for the capital flow of international financial institutions can significantly improve social welfare.This paper gives the following policy recommendations: Firstly, we should increase the significance of external financial risk prevention. In recent years, the rapid growth of international capital flows has brought new challenges to new economies. Secondly, we should strengthen the supervision of cross-border capital flows to resist external financial risks. In the process of further opening up of China’s financial market, the People’s Bank of China should improve the macro-prudential supervision of foreign exchange market, real estate market and credit market to prevent systemic financial risks. Finally, we should properly solve the financing problems of enterprises. For China, capital outflows will aggravate the already severe credit market pressure. Therefore, on the one hand, resist the external financial risk contagion through the macro-prudential supervision mechanism of capital flows; on the other hand, maintain the liquidity at a reasonable level through the flexible operation of monetary policy, in order to support economic growth.

Under ordinary circumstances, the fiscal implications of central bank policies tend to be seen as relatively minor and escape close scrutiny. The global financial crisis of 2008, however, demanded an extraordinary response by central banks which brought to light the immense power of central bank balance sheet policies as well as their major fiscal implications. Once the zero lower bound on interest rates is reached, expanding a central bank's balance sheet becomes the central instrument for providing additional monetary policy accommodation. However, with interest rates near zero, the line separating fiscal and monetary policy is blurred. Furthermore, discretionary decisions associated with asset purchases and liquidity provision, as well as with lender-of-last-resort operations benefiting private entities, can have major distributional effects that are ordinarily associated with fiscal policy. In the euro area, discretionary central bank decisions can have immense distributional effects across member states. However, decisions of this nature are incompatible with the role of unelected officials in democratic societies. Drawing on the response to the crisis by the Federal Reserve and the ECB, this paper explores the tensions arising from central bank balance sheet policies and addresses pertinent questions about the governance and accountability of independent central banks in a democratic society.

PurposeWhat causes the downward trend of real interest rates in major developed economies since the 1980s? What are the challenges of the near-zero interest and inflation rates for monetary policy? What can the policymakers learn from the latest developments in the monetary and interest rate theory? This paper aims to answer these questions by reviewing both basic principles of interest rate determination and recent academic and policy debates.Design/methodology/approachThe paper critically reviews the explanations for the downward trend of real interest rates in recent decades and monetary policy options in a near-zero interest rate environment.FindingsThe decline of real interest rates is likely an outcome of multiple technological, social and economic factors including diminished productivity growth, changing demographics, elevated tail-risk concerns, time-varying convenience yields of safe assets, increased global demand for safe assets, rising wealth and income inequality, falling relative price of capital, accommodative monetary policies, and changes in industry structure that alter the investment and saving behaviors of the corporate sector. The near-zero interest rate limits the space of central banks' response to economic crises. It also challenges some conventional wisdoms of monetary theory and sparks radically new ideas about monetary policy.Originality/valueThis survey differs from the existing work by taking a broader view of both economics and finance literature. It critically assesses the economic forces driving the global decline of real interest rates through the lens of basic principles and empirical evidence and discusses the merits and limitations of each proposed explanation. The study emphasizes the importance of a better understanding of economic forces driving diverging trends of corporate investment and saving behaviors. It also discusses the implications of the neo-Fisherism and the fiscal theory of price level for monetary policy in a low interest rate environment.

In computer science, reinforcement learning is a powerful framework with which artificial agents can learn to maximize their performance for any given Markov decision process (MDP). Advances over the last decade, in combination with deep neural networks, have enjoyed performance advantages over humans in many difficult task settings. However, such frameworks perform far less favorably when evaluated in their ability to generalize or transfer representations across different tasks. Existing algorithms that facilitate transfer typically are limited to cases in which the transition function or the optimal policy is portable to new contexts, but achieving “deep transfer” characteristic of human behavior has been elusive. Such transfer typically requires discovery of abstractions that permit analogical reuse of previously learned representations to superficially distinct tasks. Here, we demonstrate that abstractions that minimize error in predictions of reward outcomes generalize across tasks with different transition and reward functions. Such reward-predictive representations compress the state space of a task into a lower dimensional representation by combining states that are equivalent in terms of both the transition and reward functions. Because only state equivalences are considered, the resulting state representation is not tied to the transition and reward functions themselves and thus generalizes across tasks with different reward and transition functions. These results contrast with those using abstractions that myopically maximize reward in any given MDP and motivate further experiments in humans and animals to investigate if neural and cognitive systems involved in state representation perform abstractions that facilitate such equivalence relations.

Die durch die grose Finanzkrise induzierten hohen Schuldenquoten zwingen die Staaten zu Masnahmen des Schuldenabbaus. Financial Repression bietet dafur Moglichkeiten. Sie entsteht, wenn staatliche Instanzen den Finanzmarkten Masnahmen oktroyieren und manifestiert sich in erster Linie in niedrigen Realzinsen. Financial Repression wird als weiteres Charakteristikum einer „neuen Normalitat“ der Weltwirtschaft mit niedrigen Potenzialwachstumsraten und Zinssatzen sowie hohen staatlichen Schuldenstanden angesehen. Der Beitrag stellt dieses Charakteristikum vor und beurteilt seine empirische Relevanz.

The Austerity Paradox: I See Austerity Everywhere, But Not in the Statistics