CS/CMI/WMEMAFS/38/2013

Page | 1

Distr.

LIMITED

CS/CMI/WMEMAFS/38/2013

August 2013

Original: ENGLISH

COMMON MARKET FOR EASTERN

AND SOUTHERN AFRICA

Regional Workshop on Macro-econometric Modeling

for Assessing Financial System Stability

12-16 August, 2013

Nairobi, Kenya

REPORT ONREGIONAL WORKSHOP ON MACRO-ECONOMETRIC

MODELLING FOR ASSESSING FINANCIAL SYSTEM STABILITY

2013 (IZ-mkc)

  1. INTRODUCTION
  1. The 18th meeting of the COMESA Committee of Central Bank Governors which was held in Kigali, Rwanda in December 2012 assigned COMESA Monetary Institute (CMI) to undertake a workshop on Macro-econometric Modeling for Assessing Financial System Stability.
  2. The core objectives of the workshop are:

i)Build capacity of member countries to enable them to implement the COMESA Framework for Assessing Financial System Stability; and

ii)Contribute to knowledge sharing and networking between member States on issues related to financial system stability

  1. CMI,therefore, organized the workshop from 12-16 August 2013 in Nairobi, Kenya
  1. ATTENDANCE, OPENING OF THE WORKSHOP, ADOPTION OF THE AGENDA AND ORGANISATION PF WORK
  1. The workshop was attended by delegates from Central Banks of Burundi, Congo (DR), Egypt; Kenya, Madagascar, Malawi, Rwanda, Sudan, Swaziland, Uganda and Zambia. The list of participants is contained in Annex II..

Opening of the Workshop (Agenda item i)

  1. Mr. Ibrahim A. Zeidy, Director of the Institute, in his opening statement underscored that soundness of the financial system is a key part of the infrastructure for strong macroeconomic performance which is crucial for enhanced financial integration and economic welfare of member countries. He also stated that achieving diversification and stability in the financial system is as well crucial for the region, since COMESA is looking forward to trade in financial services and cross border banking. He pointed out that cross border banking poses a major challenge to supervisors in the region, as regards the need for close cooperation and sharing supervision information, both at home and across borders and sharing responsibility.
  2. He expressed his firm belief that the workshop will be an important input to the capacity efforts of member countriesto independently conduct financial stability assessment at country level.
  3. Professor Kinandu, Muragu, Executive Director of the Kenya School of Monetary Studies also made a statement. In his statement, he expressed his confidencethat the workshop will enhance the capacity of member countries and enable them to successfully implement the COMESA Framework for Assessing Financial System Stability. He pointed out that the COMESA Assessment Framework for Assessing Financial System Stability is a useful tool to gauge the resilience of the financial system in the region to endogenous and exogenous shocks as well as to craft appropriate risk mitigation measures for financial system stability.

Adoption of the Agenda and Organisation of Work (Agenda item ii)

  1. The workshop adopted the following agenda items:

i)Opening of the workshop

ii)Adoption of the agenda and organization of work;

iii)Role, Applicability & Approaches to Macro-econometric Modelling for Financial System Stability;

iv)Introduction to Econometrics of Macro Financial Modelling;

v)Characterising financial cycles and business cycles, and configuring countercyclical capital buffer rules;

vi)Longrun equation estimation and dynamic analysis;

vii)Modelling macro-financial linkages using single equation models and multivariate Cointegration equations

viii)Any other Business

ix)Wrap up session

  1. The workshop agreed on the work programme which is contained in Annex 1 of this report. The workshop also agreed on the following hours of work:

Morning: 9.00-12.30 hours

Afternoon: 14.00-17.00hrs

  1. ACCOUNT OF PROCEEDINGS

Overview of the COMESA Financial System Stability Assessment Framework (Agenda item 3)

  1. Mr. Gift Chirozva made a presentation under this agenda item. In his presentation he pointed out that financial stability is often characterised as a multi-dimensional and multifaceted phenomenon prone to various internal and external shocks. He also pointed out that a number of COMESA member countries produce financial stability reportslacking in overall financial stability assessment, forward looking outlook and granularity, thereby rendering comparison of financial stability assessment over time within a given country and /or across nations, a daunting task.
  2. He underscored that the COMESA Financial Stability Assessment Framework provides for systematic monitoring of the individual parts of the financial system (institutions, markets, and infrastructure); components of the real economy (households, firms, and public sector); global macro financial developments; and event risk (e.g. catastrophes.) He also pointed out that the framework provides a structured, comprehensive and conceptually sound analytical framework for the assessment and measurement of systemic stability over time and across nations, via the SHIELD rating system.
  3. He emphasised that the SHIELDS rating system provides structure and discipline to the process of continuous information gathering, analysis, and monitoring; as well as formulation of risk mitigation strategies. He stated that by construction the SHIELDS rating system can accommodate any theoretical and empirical advances, judgemental and professional insights as part of the overall assessment of financial system stability.
  4. He also pointed out that the COMESA Framework for Financial Stability Assessment relies, for its inputs, on concepts and international standards that are well established in the microprudential supervision fraternity. The SHIELDS rating system, like CAMELS, is based on a comprehensive quantitative and qualitative evaluation criteria.
  5. Following the presentation the workshop noted that while there are a proliferation of research in systemic risk and Financial Stability Assessments, the concepts are still evolving; hence there is always room for improvement and learning from each other’s experiences.
  6. The workshop also noted the adoption of the scoring and rating methodologies proposed in the draft handbook will facilitate implementation of the COMESA Financial Stability Assessment Framework Stability in identifying, analyzing, measuring , mitigating, monitoring and reporting financial stability issues, risks and vulnerabilities
  7. Regarding quality assurance issues related to the assignment of SHIELDS ratings, it was noted that the involvement of a Financial Stability Committee by whatever name it might be called, would be necessary to ensure buy-in, transparency and accountability. It was further noted that there is no one best solution regarding the institutional architecture.

Recommendations

  1. The workshop recommended the following:

i)There is still room for great improvement toward implementation of the COMESA

Framework for Financial System Stability, by way of aggressive in-country

workshops and practical missions, to ensure that a wide spectrum of

stakeholders are adequately exposed to the framework and rating

methodologies.

ii)The CMI should conduct 2014 hands on training on SHIELDS ratings based on a fullfledged case study and develop a stylised sample solution.

Role, Applicability & Approaches to Macro econometric Modelling for Financial Stability (Agenda item iv)

  1. Under this agenda item, Mr. Gift Chirozva pointed out that in practice financial stability is a complex phenomenon dependable on iterartive interactions of many risks and feed back effects from financial stability to the real economy. As a result most academics and practitioners concurthat there will never be one best model for all purposes, hence the need for a suite of models.
  2. He informed the workshop that the use of on-going macro econometric modelsis subject to a lot of debate since the global financial crisis. He said that despite of the debate, macro-econometric modelling helps to inform understanding of the empirical state of systemic risk measurementand can play the following important roles
  3. Provide a macro econometric based criterion to evaluate the success of alternatives of systemic risk measures;
  4. The methodologies also facilitate evaluation of the validity of many proposed systemic risk measures,identification of leading, lagging, and coincident indicators;
  5. Assessment of the real economic activity effects of financial stability;
  6. Provide a rigorous approach that aggregates information about financial sector stress and challenges to the real sector activity using multivariate methodologies/approaches.
  7. Provides an Early warning system (EWS) from a large cross section of “risk predictor” variables that contain information regarding the variable of interest.
  8. Helps inform interpretation of systemic risk measurements.
  9. The workshop considered the role and applicability of various approaches to macroeconomic modelling, ranging from those that are high in theoretical coherence to those high in data coherence on the Pagan model classification spectrum.
  10. It was noted that models used for financial stability purposes should take into account certain characteristics that distinguish them from the classical economic models, including contagion, default, heterogeneity and missing markets.

Introduction to Econometrics of Macro Financial Modelling (Agenda item v)

  1. Mr. TinasheBvirindi made a presentation under this agenda item. In his presentation he discussed the following:

i)Basics of Ordinary Least(OLS) Square regressions

ii)OLS assumptions and parameter estimation;

iii)Formulating and testing hypothesis;

iv)Loading data to Eviews

v)Data transformation and manipulation; and

vi)Running basic regressions;

Characterising financial cycles and business cycles, and configuring countercyclical capital buffer rules (Agenda item vi)

  1. Under this agenda item Mr. Bvirindi discussed the implications of violations of CLR assumptions He emphasised that parameter estimates needed to meet the Gauss Markov theorem assumptions for them to be best linear unbiased estimator of the true population regression function. It was emphasised that estimatedparameters need to satisfy the desirable properties of efficiency, consistency and unbiasedness. He underscored that models that violated these primary assumptions would often fail to produce reliable forecasts.
  2. He explained that current macroprudential policy debates and the recent efforts of the Basel committee have been directed towards the development and use of macroprudential tools that reduce procyclicality. In the recent Basel III requirement, the committee identified procyclical capital buffers as important tools to curb procyclicality and suggested that the buffer requirement should be configured on the credit to GDP ratio. He demonstrated to the workshop through a hands on exercise, the use of filtering techniques such as the Hodrick and Prescott filter, the Baxter and King filter as well as Christiano and Fitzgerald band pass filters in the computation of the excess credit to GDP ratio for the configuration of the countercyclical capital buffer andfor the characterisation of the financial cycle and the business cycle.
  3. He emphasised that characterisations of the financial cycle and the business cycle may differ from country to country such that variable that may best serve as indicators for the build-up of the countercyclical capital buffer in the build-up phase of vulnerability may differ from country to country.

Recommendation

  1. Countries should identify lead indicators that can be used in configuring the countercyclical capital buffers. The identification of the lead indicators should be informed by country specific studies as indicators that are best suited as lead indicators for the build-up of the buffer may not necessarily be the best indicators for the release of the capital buffer.

Longrun equation estimation and dynamic analysis (Agenda item vii)

  1. Mr. Bvirindi informed the workshop that theories in economic and finance theory and based on long run relationships. He explained that regressions that are based only on basic regressions fail to capturelongrun information in the data series. He also explained techniques such as Cointegration which are necessary for capturing longrun relationships between variables and the shortrun dynamics of adjustment to equilibrium in the event of disturbances in the system. He said, the technique helps to separate permanent effects from transitory effects in the model.
  2. During this session, participants estimated single Cointegration equations and performed diagnostic tests on the models using Eviews. These excercises revealed to the participants that the model performance is compromised once assumptions of residuals are violated. And that once a model has failed diagnostic tests, it cannot be used for policy simulations and forecasting.

Modelling Macro-Financial Linkages using Single Equation Models and Multivariate Cointegration equations (Agenda item viii)

  1. In this session, Mr. Bvrindi explained that, in practise financial stability analysis is a complex phenomenon dependable on the iterative interactions of many risks and feedback effects between the financial system and the macro-economy. He said these multiple and complex interactions require a suit of models for their estimation. He stated that Pagan (2003) specifies a range of models that are currently in use in modelling in central banks. He identifies data coherent models such as Vector Auto Regression (VAR) and Dynamic Aggregative Models as forming part of the suit of models currently in use in central banks.
  2. He demonstrated to the workshop, the modelling of macro financial linkages using a dynamic aggregative model. He also explained the estimation of a three equation model capturing asset price dynamics, credit market dynamics and money demand.
  3. During this session participants conducted shock simulations on the model in question to demonstrate that shocks in variables in one part of the system may be amplified or attenuated through endogenous interactions between the variables in the system, which may amplify or attenuate risks. Furthermore, participants performed in-sample and out of sample forecasts using the model and then computed the impulse response functions to trace the time profile of shocks on the system. Such excerciseallows the central bank to tell in advance how shocks may mutate in the system and thus provides them lead time for policy responses.
  4. He emphasised the importance of multivariate analysis through multivariate cointegration equations. He said that, the use of multivariate equations allows one to understand the permanent effect and temporary effects of shocks to the system in more complete ways and provides for a richer study of the dynamics of adjustment. He explained that, these techniques will enable countries to tell the relative importance of different variables in explaining shocks to their systems and to judge the magnitudes of observed variability to target variables that is explained by the identified explaining variables.
  5. In the discussions that followed this session, the workshop was fully convinced that with the rest of the world moving towards the estimation of DSGE models for their economies, it is advisable for countries not to lag far behind in the adoption of techniques for macro-financial modelling. Given the data challenges that face most African countries, the use of relatively simple estimation techniques which need less data such as structural VAR models was, therefore recommended.

Wrap-up session

  1. The purpose of this session was to get feedback from participants on the relevance and usefulness of the workshop. The participants appreciated the workshop and fully agreed that it enabled them to: i) enhance their understanding of modelling financial stability and provided hands on training on empirical modelling financial stability; ii) Contribute to knowledge sharing and networking between member states on the designing and using econometrics for analysing financial stability.
  1. In the discussions that followed the participants felt the need for COMESA Monetary Institute to organise such Workshop/training on a yearly basis.

Recommendation

  1. The participants of the workshop recommended the following:

i)CMI should continue to conduct hands on training on financial system stability assessments and econometric modelling of financial stability, including macro stress testing with hypothetical data.

ii)CMI should organise a workshop 2014 on Macrofinancial modelling using structural VAR