DETERMINANTS OF LOAN LOSS PROVISIONS OF VIETNAMESE COMMERCIAL BANKS

TABLE OF CONTENT SUPERVISOR’S REMARK ABSTRACT ACKNOWLEDGEMENT LIST OF ABBREVIATIONS LIST OF TABLES AND FIGURES Chapter 1: INTRODUCTION .................................................................................1 1.1. The necessity of the research............................................................................1 1.2. The overview of the research............................................................................1 1.3. The purpose of the research..............................................................................3 1.4. The object and scope of the research................................................................3 1.5. The methods of the research .............................................................................4 1.6. The novelty and contribution of the research ...................................................4 1.7. The structure of the research.............................................................................5 Chapter 2: LITERATURE REVIEW .....................................................................6 2.1. Credit risk and loan loss provisions of commercial banks ...............................6 2.1.1. Credit risk ...................................................................................................6 2.1.2. Loan loss provision ....................................................................................8 2.2. Fundamental literature on determinants of loan loss provisions ....................15 2.2.1. Bankspecific factors................................................................................15 2.2.2. Macroeconomic factors ............................................................................24 Chapter 3: METHODOLOGY ..............................................................................27 3.1. Data and sample..............................................................................................27 3.2. Hypothesis development.................................................................................28 3.3. Variables ........................................................................................................ 28 3.3.1. Dependent variable.................................................................................. 29 3.3.2. Independent variables.............................................................................. 29 3.4. Econometric model ........................................................................................ 35 Chapter 4: DATA ANALYSIS AND RESULTS................................................. 38 4.1. Introduction of Vietnamese commercial banks ............................................. 38 4.3. Regression results .......................................................................................... 44 4.3.1. Descriptive statistics and correlation matrix ........................................... 44 4.3.2. Empirical results...................................................................................... 49 Chapter 5: THE OUTLOOK, ORIENTED DEVELOPMENT AND RECOMMENDATIONS FOR IMPROVEMENT OF LOAN LOSS PROVISIONS OF VIETNAMESE COMMERCIAL BANKS ......................... 59 5.1. The outlook of Vietnamese commercial banks ............................................. 59 5.2. Orientation for loan loss provisions of Vietnamese commercial banks ........ 60 5.3. Recommendations for improvement of loan loss provisions of Vietnamese commercial banks ................................................................................................. 62 5.3.1. Recommendations for commercial banks ............................................... 62 5.3.2. Recommendations for the government.................................................... 70 CONCLUSION....................................................................................................... 78 REFERENCES ....................................................................................................... 79 ANNEX 1 ANNEX 2

HO CHI MINH CITY CAMPUS

GRADUATION THESIS

Major: International Business Economics

DETERMINANTS OF LOAN LOSS PROVISIONS OF

VIETNAMESE COMMERCIAL BANKS

Author: Pham Thi Thanh Yen Student ID: 1201016673

Class: K51CLC1 Course: K51 Academic Supervisor: MA Nguyen Thu Hang

Ho Chi Minh City, May 2016

Thesis ID: 69

TRƯỜNG ĐẠI HỌC NGOẠI THƯƠNG

CƠ SỞ II TẠI TP HỒ CHÍ MINH NHẬN XÉT KHÓA LUẬN TỐT NGHIỆP

Họ và tên sinh viên: Phạm Thị Thanh Yến MSSV: 1201016673 Tên đề tài: DETERMINANTS OF LOAN LOSS PROVISIONS OF

VIETNAMESE COMMERCIAL BANKS Điểm tinh thần, thái độ, chuyên cần (tối đa 1 điểm; cho điểm lẻ đến 0,1):

Ý kiến nhận xét (khoanh tròn lựa chọn phù hợp):

1 Sinh viên đã nghiêm túc thực hiện KLTN theo sự hướng dẫn của GVHD GVHD chịu trách nhiệm về tên đề tài, mục đích, đối tượng, phạm vi & phương pháp nghiên cứu và tên các chương, các đề mục chi tiết (3 chữ số)

2 Sinh viên đã thực hiện theo hướng dẫn của GVHD nhưng chưa đầy đủ GVHD chịu trách nhiệm về tên đề tài, mục đích, đối tượng, phạm vi, phương pháp nghiên cứu và tên các chương, các đề mục chính (2 chữ số)

3 Sinh viên không thực hiện đầy đủ hướng dẫn của GVHD GVHD không chịu trách nhiệm về đề tài

4 Sinh viên không thực hiện hướng dẫn của GVHD GVHD không đồng ý cho sinh viên nộp KLTN

Tp Hồ Chí Minh, ngày … tháng … năm 2016

Giảng viên hướng dẫn

(Ký và ghi rõ họ tên)

The thesis explores the determinants of loan loss provisions of Vietnamese commercial banks including income smoothing, signaling, capital management, pro-cyclical behaviors and other bank-specific factors It does this by looking at a sample

of 39 Vietnamese commercial banks over the period 2007 to 2014 By examining the impacts of earlier mentioned factors, this thesis contributes to the literature on how Vietnamese commercial banks use loan loss provisions and why they are set up at a certain level

The findings demonstrate that loan loss provisions of commercial banks in Vietnam are in significant relationship with income smoothing, signaling, pro-cyclicality and some other bank-specific factors comprising nonperforming loans, loan growth and one-year ahead provisions Meanwhile, for the conjectures that Vietnamese commercial banks engage in capital management through the manipulation of loan loss provisions and provisions have any significant association with bank size, there

is no sufficient evidence to support these suggestions

The author wants to give best regards and sincere thanks to the author’s supervisor

Ms Nguyen Thu Hang for her helpful advices and useful guide throughout the time completing this thesis The author also wants to express great gratitude to all the teachers at Foreign Trade University – Ho Chi Minh City campus for providing and teaching the author with specialized and useful knowledge which stands as a strong basis for this thesis content

Despite special drives, certain shortcomings are inevitable throughout this thesis due

to strict format regulations together with the lack of practical and specialized experience of the author The author sincerely respects sympathy and concessions in assessment of this thesis

Ho Chi Minh City, May 2016

Pham Thi Thanh Yen

SUPERVISOR’S REMARK

ABSTRACT

ACKNOWLEDGEMENT

LIST OF ABBREVIATIONS

LIST OF TABLES AND FIGURES

Chapter 1: INTRODUCTION 1

1.1 The necessity of the research 1

1.2 The overview of the research 1

1.3 The purpose of the research 3

1.4 The object and scope of the research 3

1.5 The methods of the research 4

1.6 The novelty and contribution of the research 4

1.7 The structure of the research 5

Chapter 2: LITERATURE REVIEW 6

2.1 Credit risk and loan loss provisions of commercial banks 6

2.1.1 Credit risk 6

2.1.2 Loan loss provision 8

2.2 Fundamental literature on determinants of loan loss provisions 15

2.2.1 Bank-specific factors 15

2.2.2 Macroeconomic factors 24

Chapter 3: METHODOLOGY 27

3.1 Data and sample 27

3.2 Hypothesis development 28

3.3.1 Dependent variable 29

3.3.2 Independent variables 29

3.4 Econometric model 35

Chapter 4: DATA ANALYSIS AND RESULTS 38

4.1 Introduction of Vietnamese commercial banks 38

4.3 Regression results 44

4.3.1 Descriptive statistics and correlation matrix 44

4.3.2 Empirical results 49

Chapter 5: THE OUTLOOK, ORIENTED DEVELOPMENT AND RECOMMENDATIONS FOR IMPROVEMENT OF LOAN LOSS PROVISIONS OF VIETNAMESE COMMERCIAL BANKS 59

5.1 The outlook of Vietnamese commercial banks 59

5.2 Orientation for loan loss provisions of Vietnamese commercial banks 60

5.3 Recommendations for improvement of loan loss provisions of Vietnamese commercial banks 62

5.3.1 Recommendations for commercial banks 62

5.3.2 Recommendations for the government 70

CONCLUSION 78

REFERENCES 79 ANNEX 1

ANNEX 2

Abbreviation Full word

CDO Collateralized Debt Obligations

CFA Chartered Financial Analyst

EU European Union

FEM Fixed Effects Model

GDP Gross Domestic Product

GMM Generalized Method of Moments

IFRS International Financial Reporting Standards

M&A Merger and Acquisition

OECD Organization for Economic Cooperation and Development OLS Ordinary Least Squares

PDF Probability Density Function

REM Random Effects Model

SBV State Bank of Vietnam

TPP Trans-Pacific Partnership

USD United States Dollar

US United States

VAMC Vietnam Asset Management Company

VAS Vietnamese Accounting Standards

WTO World Trade Organization

Tables and Figures Page Figure 2.1: The illustration of loan loss provisions and loan loss allowance

in the bank’s financial statements

10

Figure 2.2: Probability density function of losses, unexpected and expected

losses, and economic capital

Figure 4.3: The rating of nonperforming loan ratios of Vietnamese banking

system of different entities from 6/2011 to 2/2013

Table 4.2: Pairwise correlation matrix between variables 47

Table 4.4: Estimation of loan loss provision determinants – OLS, FEM,

GMM

50

Table 4.5: Coefficients of different combinations of independent variables

from different multivariable regressions on loan loss provisions

57

Chapter 1: INTRODUCTION

1.1 The necessity of the research

Borrowing and lending are the two most important activities of commercial banks, accounting for most of their income When there is an increase in these activities or

a credit growth, banks’ earnings from the difference between depositing and lending interests will simultaneously rise However, credit growth also accompanies the increase in nonperforming or bad loans as a result of reckless lending In order to alleviate the ramifications of bad loans, three common solutions for Vietnamese banks are selling them to Vietnam Asset Management Company (VAMC), utilizing loan loss provisions, and attempting to recover as many bad debts as possible While bad debts must satisfy five strict requirements to be purchased by VAMC, their recovery by banks’ attempts is greatly dependent on borrowers Therefore, loan loss provisioning becomes the most active solution for banks Nonetheless, besides the ability to lessen the consequences of bad loans, loan loss provisions of high amount will result in a considerable loss in banks’ profit as well as the government revenue Thus, the research on the determinants of loan loss provisions is very necessary for building the most suitable provision amount In addition, there have been numerous studies on loan loss provisions in many countries around the world but not many in Vietnam Therefore, the author begins this research

1.2 The overview of the research

There have been a lot of researches on loan loss provisions worldwide, providing fully both theoretical and empirical premise for application and further studies However, the most common approach to components of loan loss provision is based

on the discretion of bank managers On that account, loan loss provision determinants are classified into discretionary and non-discretionary ones

Discretionary components are caused by management’s use of loan loss provisioning for its own objectives (Craigwell and Elliott, 2011) which can be income smoothing, capital management, and signaling These are the three most important issues in determining the loan loss provisions by banks and closely related with each other Banks are motivated to manage the variability of reported earnings, manage the

capital adequacy ratios required by the regulators as well as signal the future performance of the banks for public attention (Hasnie et al., 2015) Bouvatier and Lepetit (2008), Pinho and Martins (2009), Floro (2010), Bushman and Williams (2012), Norden and Stoian (2013), and Curcio and Hasan (2013) also posit the banks’ use of loan loss provisions for income smoothing and capital management in their researches

In terms of non-discretionary components which is beyond banks’ control, loan loss provisions are susceptible to pro-cyclicality Pro-cyclicality is the phenomenon amplifying feedbacks within the financial system and between the financial system and the macro-economy (Valverde and Fernandez, 2014) In relation to this, Laeven and Majoni (2003) and Bikker and Metzemakers (2005) show that provisioning behavior is related to the business cycle Another empirical evidence can be found for UK (Pain, 2003), Hong Kong and Singapore (Eng and Nabar, 2007), France (Dinamona, 2008), Indonesia (Suhartono, 2012), and Malaysia (Adzis et al., 2015)

In Vietnam, there have been some writings about loan loss provisions Nguyen Thi Lan Phuong (2015) shows the current conditions and gives some recommendations for improving loan category and loan loss provisions at Vietnam Maritime Commercial Joint Stock Bank by way of evaluating the lessons from France, America, England, Singapore, and China In her master thesis in 2013, Nguyen Thi Diem Kieu analyzes some determinants of loan loss provisions including GDP growth, bank size, credit growth, credit risk, interest rate, earnings before taxes and provisions, and bad loans by means of pooled ordinary least squares (OLS) and random effects model (REM) The most recent study is “An analysis of loan loss provisioning behavior in Vietnamese banking” written by Bryce et al (2015) published on Finance Research Letters Volume 14, page 69-75 This paper investigates loan loss provisioning by Vietnamese banks during the period 2006–

2012 by testing the capital, income smoothing and cyclical management hypotheses and examining whether the inclusion of X-efficiencies and/or risk control variables influences provisioning When the X-efficiency estimates are incorporated into the

models, Vietnamese banks do not exhibit counter-cyclical or capital management manipulation by managers, but counter-cyclical income smoothing

At Foreign Trade University Ho Chi Minh City Campus, this thesis is the first of its kind analyzing the determinants of loan loss provisions The author uses the approach

of abovementioned foreign studies by Generalized Method of Moments (GMM) model

Although international economists and practitioners have paid a lot of attention to loan loss provisions, they have not yet been studied thoroughly in Vietnam On that account, the author begins this thesis with hope that it can give some contributions to the groundwork of Vietnamese loan loss provisioning

1.3 The purpose of the research

The thesis aims at establishing and analyzing determinants of loan loss provisions of Vietnamese commercial banks The purpose is to provide an insightful approach toward loan loss provisions and independent factors that can affect them significantly Based on empirical results of running econometric regression models, the impact of bank-specific and macro-economic factors on loan loss provisions is explained in depth, and some economic implications are pointed out so that these influences can

be explained and fully understood from economic viewpoint The results can later be used by both banks’ managers and government officials for controlling and regulating financial institutions

1.4 The object and scope of the research

The research object is loan loss provisions of Vietnamese commercial banks and the bank-specific as well as macro-economic factors that are expected to have impact on loan loss provisions in regression models

The research period is from 1/1/2007 to 31/12/2014 This period includes both stages

in and after the global financial crisis 2007-2009 in order to examine the economic influence on loan loss provisions As the financial climate is very unstable and the banking system in Vietnam is being reformed, the model’s predictability is just in short term (one year)

macro-The scope of the research is limited to state-owned commercial banks and joint stock commercial banks in Vietnam with different annual totals The number of banks is based on their audited financial reports published every year Consequently, there are

312 observations over the period of eight years

1.5 The methods of the research

-Collecting papers written by many international economists and practitioners regarding the topic of loan loss provisions From those sources, the author applies the method of analysis, synthesis and comparison to draw out the most suitable model to

be used to assess the determinants of loan loss provisions of commercial banks in the context of Vietnam

-Collecting bank data from annual financial reports of Vietnamese commercial banks

As the banking system of Vietnam still has many problems unsolved, the number of commercial banks every year is not alike Hence, the author only uses the complete annual financial reports published on the reliable website - http://finance.vietstock.vn/ Vietnamese economic data is collected from World Bank database

-Using econometric regression model based on the theory of GMM developed by Arellano and Bond (1991), Arellano and Bover (1995), and Blundell and Bond (1998)

to analyze and evaluate the relationship between the affecting factors and loan loss provisions at different significance levels in STATA 12

1.6 The novelty and contribution of the research

The majority of researches on Vietnamese commercial banks relating to loan loss provisions just focus on credit risk, classification of bad loans and how banks’ managers and government officials should use loan loss provisions to solve them In this thesis, the time scope is relatively longer than other studies in the context of Vietnam in order to test the influence of business cycle on loan loss provisions With

an unbalanced panel and lagged independent variable, the regression models employed in the thesis are based on GMM On the findings, the author concludes the significant motives behind provisioning decisions of Vietnamese commercial banks and give some suggestions for improvement

1.7 The structure of the research

The structure of the thesis consists of five following parts:

Chapter 1: Introduction

Chapter 1 introduces some general information about the thesis such as the necessity, overview, purpose, object and scope, method, novelty and contribution of the research The structure of the thesis is also mentioned in this chapter

Chapter 2: Literature review

In Chapter 2, fundamental knowledge relating to credit risk and loan loss provision are reviewed Besides, the author also lists the potential determinants of loan loss provisions and the related findings of different researchers

Chapter 3: Methodology

This chapter describes the method of regression model formation in order to examine the impact of different factors on loan loss provisions of Vietnamese commercial bank The choice of the variables and their economic implications are also explained

in this chapter

Chapter 4: Data analysis and results

The first part of this chapter summaries the current operating conditions of Vietnamese commercial banks together with their loan loss provisioning activities during 2007-2014 Then, the results derived in Chapter 3 are presented and explained thoroughly in the context of Vietnamese banking system

Chapter 5: The outlook, oriented development and recommendations for improvement of loan loss provisions of Vietnamese commercial banks

In the final chapter, the outlook and development orientation are described according

to the analyzed results in Chapter 4 Some recommendations for improving loan loss provisioning of Vietnamese commercial banks are also discussed for better future operation

Chapter 2: LITERATURE REVIEW

2.1 Credit risk and loan loss provisions of commercial banks

2.1.1 Credit risk

2.1.1.1 Definition

Credit risk is the risk that the counterparty will not pay when a positive amount

is owed at settlement The larger is the value to one party , the greater the credit risk to that party (2013, CFA Level 2, book 5) In banking system, where the primary business is the collection and investment of depositors’ funds, banks bear credit risk, the possibility that the borrower will fail to repay a loan or meet a contractual obligation as promised (Walter, 1991) Credit risk is noted to be the most significant

of all the risks that banks must manage as it can result in potential losses and failure

There is a wide range of indicators available in the market used to evaluate bank credit risk According to Fitch Solutions, they include traditional bank credit ratings, credit-default swap implied ratings and spreads, fundamental financial data, and a suite of market-based indicators Among these, financial data such as the ratio of nonperforming loans and the ratio of total loans to total assets are the most commonly used by researchers on topic relating to bank credit risk

According to Suhartono (2012), credit risk plays an important role in bank risk management since banks always try to develop their own credit risk models with the aim of increasing bank portfolio quality As summarized by the Federal Reserve System Task Force on Internal Credit Risk Models (1998) and the Basel Committee

on Banking Supervision (1999), there are a wide variety of credit risk models that differ in their underlying assumptions, such as their definition of credit losses, e.g default models define credit losses as loan defaults, while mark-to-market or multi-state models define credit losses as ratings migrations of any magnitude However, the common motive of these models is to predict the probability distribution function

of losses that may arise from a bank’s credit portfolio (Lopez and Saidenberg, 1999) Thanks to these models, banks can greatly improve their risk management capabilities on deciding how best to control the credit risk in a portfolio, such as by setting up the appropriate loan loss reserves or by selling loans to reduce risk

- Credit spread risk refers to the fact that the default risk premium required in the market for a given rating can rise, even while the yield on Treasury securities of similar maturity remains unchanged An increase in this credit spread

raises the required yield and reduces the price of a bond Spreads tend to widen in

poor performing economies

- Downgrade risk is the risk that a credit rating agency will lower a bond's

rating These agencies, such as Moody's, Standard & Poor's and Fitch, give a bond issuer a rating that indicates the possibility of default The ratings range from the best rating AAA to AA, A, BBB These are the ratings for investment-grade bonds Once bonds fall into the BB, B, CCC ranges they become junk bonds or high yield securities If one of these rating agencies downgrades a company's rating, the market value of the company’s bonds will typically decrease

2.1.1.3 Cause and effect

Credit risk arises as soon as a loan is granted to a borrower and the bank is compensated for assuming credit risk by means of interest payments from the borrower or issuer of a debt obligation Credit risk implies in all bank lending activities but at different levels The loans of higher risk levels usually gain more profits but expose to more potential losses at the same time On that account, banking institutions and bank regulators should closely monitor lending activities as poor monitoring in loan activity may lead to bank failure (Adzis et al., 2015)

However, Pain (2003) suggests that in periods of economic boom, banks are inclined

to take on greater risks, underestimate future losses because lending criteria are relaxed or concentrations on loan exposures increase At this time, an accumulation

of potential risk (expected losses) is built up with high loan growth rate at the expense

of declining credit quality During subsequent economic downturns, this over-lending leads to a sharp increase in bad debts as many low-quality borrowers cannot service their debts As this may occur when bank income is weaker due to slower loan demand growth, such losses can actually reduce bank’s existing capital Further, during such recession periods, bank may themselves be less able to raise new capital These two reasons may lead to the violation of bank capital requirements

The cyclical feature of bank lending activities has been identified as a key ingredient

of the “inherent instability” of financial systems (Minsky, 1982) In this context, some scholars have emphasized the effects of disaster myopia that implies underestimating the possibility and magnitude of financial crises (Guttentag and Herring, 1984), herd behavior that focuses on the idea that banks’ management is preoccupied with short-term concerns and perception of reputation (Banerjee, 1992)

or the institutional memory hypothesis (Berger and Udell, 2004), stating that the capacity of bank loan officers to evaluate risk and identify potential nonperforming loans declines over time while the previous loan bust is not remembered because of loan officer turnover Empirical investigations (Asea and Blomberg, 1998; Lown and Morgan, 2006) also document the cyclicality of lending standards applied by banks and their ramifications both on the credit cycle and on the business cycle

2.1.2 Loan loss provision

2.1.2.1 Definition

Loan loss provision (or provision for credit loss) is defined as estimation for probable loan losses for the current year and this amount will be charged on the income statement as expenses on a yearly basis The purpose of this accrual expense is to absorb any losses arising from loan default by customers The reserve for loan losses (also known as allowance for possible loan losses or impaired loans) on the other hand is a contra asset account created to record loan loss provisions, accumulated

loans charged off, and any loans recoveries for the current year This account is recorded on the bank balance sheet as a deduction from total loans The increase in loan loss provisions will result in an increase in loan loss allowance and a reduction

in current net income The process of recording loan loss provisions in the income statement and loan loss reserve in the bank’s balance sheet will be illustrated more clearly and comprehensively in the next page

The setting aside of reserves reduces the provisioning institution's earnings and therefore has an adverse effect on profits While creditors reduce their risk exposure

by provisioning, they must pay a price in terms of reduced short-term profits However, neither writing down a loan nor provisioning against it reduces the debtor's contractual obligations Such relief is forthcoming only when the lending institution translates the write-down into a reduction in debt-service payments from the debtor

In that case, the bank is forgiving a proportion of the debt Ahmed et al (1998) suggests that loan loss provision is one of the main accrual expenses for commercial banks and therefore, has a significant impact on banks’ reported earnings and regulatory capital Pain (2003) also states that loan loss provision is typically one of the first quantitative indicators of loan quality deterioration and concurrently a key contributor to fluctuation in banks’ profits and capital

The presence of loan loss provisions allows banks to recognize in their profit and loss statements the estimated loss from a particular loan portfolio, even before the actual loss can be determined with accuracy and certainty (Floro, 2010) This necessarily helps banks’ directors avoid overstating their assets as well as income-earning potential and capital instead of mentioning the amount of funds lent alone Besides, data on the amount of provisions a bank holds and additions made to them are useful

to outsiders such as creditors, regulators, and investors since they provide reliable information about the risk and the quality of the loan portfolio (Walter, 1991)

Figure 2.1: The illustration of loan loss provisions and loan loss allowance in the bank’s financial statements

(Source: Adzis, 2012)

Figure 2.1 illustrates how loan loss provisions are recorded in the banks’ income statements, the example of loan loss allowance account and the recording of loan loss allowance in the banks’ balance sheets First, when loan losses are recognized or when a bank decides that some portion of a loan will not be collected and therefore must be charged off or written down, an amount of provisions will be set up to compensate for the losses This amount is then added to the loan loss reserve account

as this year’s provisions for loan losses Finally, the total ending balance of loan loss reserve is deducted from the total loans or gross loans on the asset side of a bank’s balance sheet The remaining is called net loan which is considered as the best estimate of the net realizable value of the loan portfolio as of the financial statement date (Walter, 1991)

Provisioning rules and capital requirements are linked through the coverage of credit risk: the conceptual framework of credit risk management supposes that loan loss provisions have to cope with expected losses while bank capital has to cope with unexpected ones (Bouvatier and Lepetit, 2008) Expected losses commonly occur on average and can be measured by the mean value of the frequency distribution of loan losses while unexpected losses are big but infrequent and therefore can be located far

in the tail of the frequency distribution of loan losses These two categories of shock absorbers help banks normally operate in adverse occurrences and they have really close relation as the buffer of unexpected shocks (regulatory capital) rests on the existence of the subsidiary buffer represented by the reserves created through loan loss provisions (Cavallo and Majoni, 2001) That prevailing conceptual framework is summarized in the following Figure 2.2

Figure 2.2: Probability density function of losses, unexpected and expected

losses, and economic capital

(Source: Cavallo and Majoni, 2001)

The figure describes the probability density function (PDF) of losses, unexpected and expected losses, and economic capital of a particular bank Horizontal axis represents the value of losses while vertical axis shows the probability of losses in proportion of each value in horizontal axis At point O, the loss probability equals 0 as there is no bank assuring that they will never get loss or all of their customers will service their debts perfectly At point B, the loss probability is also miniscule as a bank can hardly let itself get loss at such degree The area below PDF curve and the left of B represents 99% of the total area below PDF, which means there is averagely one bank getting excessive loss in 100 banks The probability of getting loss will get maximum value with a relatively small number and gradually decrease The area below PDF curve and the left of A is approximately 50% of total losses, namely expected losses A bank needs provisions to ensure its normal operation when this kind of loss occurs The probability of getting loss decreases from point A to point B, representing unexpected losses The buffer for these losses is capital and stipulated by regulators

on minimum amount for the whole banking system

2.1.2.2 Classification

Loan loss provisions are divided into two groups: general provisions and specific provisions According to widespread accounting practices, general provisions allude

to "ex ante" provisions and are related to future uncertain events As they refer to probabilistic losses that cannot be supported by loan specific documentation, they can

be highly judgmental, contentious, and prone to manipulation by bank managers for opportunistic reasons or for tax avoidance purposes, which may downgrade the transparency of banks and lead to negative consequences along other dimensions (Bushman and Williams, 2012) As a result, in many countries, general loan loss provisions are imposed strict accounting and regulatory restrictions on the maximum level that banks can set up For example, in some countries, accounting and regulatory restrictions mean that general provisions are limited to a small percentage of total loans General provisions are added to general reserves on liabilities Specific provisions can instead be seen as "ex-post" provisions, in which they refer to particular events (such as past due payments, or other default-like events) Specific provisions can be easily documented and are not subject to significant restrictions The amount of specific provisions depends on credit losses and it increases specific reserves, which are deducted from the asset value

Not always bank regulations refer explicitly to general or specific provisions but most

of the times regulatory requirements can be partitioned among "ex-ante" and "ex post" provisioning For instance, provisions generated by past due payments could be considered as specific provisions Provisions which are, instead, required for all loans, independently from the presence of a default event, can be considered of a general nature

In principle, provisions should be forward-looking – they should relate to future expected losses on loans However, in practice, accounting conventions in many countries mean that provisions are set in a backward rather than forward-looking manner – specific provisions can only be made once the debt is shown to have genuinely become impaired and general provisions should cover losses which have not yet been identified but which currently lie latent in the loan portfolio That is,

provisions reflect actual, rather than expected losses, which may take some time to reveal themselves (Pain, 2003)

2.1.2.3 Measurement of loan loss provision

According to Circular No 02/2013/TT-NHNN of the State Bank of Vietnam (SBV), the amount of specific provisions required to set up for each debt shall be calculated

under the following formula:

1 to Group 4 for general provision

Table 2.3 The loan loss provision ratio for the classes of debts

(Source: Circular No 02/2013/TT-NHNN)

As in the table 2.3, the loan loss provision ratio increases from Group 1 to Group 5 Group 5 is the debts that are considered can be loss Banks are required to set the loan loss provision ratio to 100% of the loan amount for debts belonging to Group 5 (Circular No 02/2013/TT-NHNN) For example, there is a customer applying a loan with the value 100,000 USD When that customer’s loan belongs to Group 2 of debt, overdue from 10 days to less than 30 days, the bank needs to reserve 5% of the loan amount This reserve amount rises step by step and when the customer’ loan is overdue above 180 days and belongs to Group 5, the bank needs to reserve 100% of the loan amount In this case, the bank needs to reserve 100,000 USD This reserving amount is used to prevent the bank from bankruptcy

2.2 Fundamental literature on determinants of loan loss provisions

2.2.1 Bank-specific factors

2.2.1.1 Income smoothing

Mulford and Comiskey (2002) define income smoothing as an earnings management technique designed to remove peaks and valleys from a normal earnings series, including steps to reduce and save profits during good years for use during bad years Despite numerous definitions concerning income smoothing, the common point is that income smoothing refers to reducing variation in incomes over time to obtain smoother income streams Bank managers normally use some accrual items in the income statements to smooth bank annual income Since loan loss provision account

is the largest accrual in commercial banks, it is used as the main instrument of earnings smoothing (Fernando and Ekanayake, 2015)

In principle, income smoothing may have a positive impact upon decreasing the cyclicality of lending In one sense, income smoothing is considered desirable, because it reduces the perceived volatility of income and thereby maintain stock price stability and result in higher firm value (Fernando and Ekanayake, 2015) Besides, income smoothing constitutes a practice that allows banks to comply with the regulatory requirements (Taktak et al., 2010) However, it may discourage bank managers from accurately disclosing loan losses, deteriorating the quality of financial reporting This will mislead outsiders such as investors, shareholders, regulators, and

tax authorities as they rely on the numbers in the financial statements To the extent that the variability of net income is a measure of risk, income smoothing may reduce bank risk perception, yet the “true” risk could be higher than the perceived risk

Numerous papers have tested the income smoothing behavior of banks through the use of loan loss provisions with contradictory results On one hand, studies by Greenawalt and Sinkey (1988) and Wahlen (1994) show that banks in the US use loan loss provisions to smooth income Additionally, Beatty et al (1995) also support the evidence of earnings management in US banks where, in particular, profitable banks use loan loss provisions to manage earnings Collins et al (1995) find a positive association between earnings management and loan loss provisions, thus supporting the notion that banks smooth income over time Bikker and Metzemakers (2005) also find statistical evidence of income smoothing in countries such as the US, France or Italy, but none in Japan, the UK or Spain More recently, Anandarajan et al (2007) show that Australian commercial banks are engaged in earnings management practices, especially if they are publicly traded By investigating the cross-country determinants of income smoothing within a sample of banks from different countries, Fonseca and Gonzalez (2008) find that the impetus to smooth earnings increases in more developed and market-oriented financial systems Moreover, according to their results, bank incentives to smooth income are lower in banking systems featured by higher levels of accounting disclosure and official or private supervision and by stricter restrictions on banking activities

Different from previous studies on income smoothing, Wetmore and Brick (1994) do not find evidence of income smoothing activities in American bank holding companies Ahmed et al (1999) explore the 1990 change in capital adequacy ratio regulations to investigate the manipulation of loan loss provisions to manage earnings, capital, and as a tool for signaling The authors find that there is no evidence

of earnings management under the new regime, but indicate that banks in the

1986-1995 period used loan loss provisions to manage regulatory capital Laeven and Majnoni (2003), Ismail et al (2005), and Taktak et al (2010) also reject the use of loan loss provisions for income smoothing in bank operation

In this thesis, following the previous literature in the examination of income smoothing behavior, the author uses the ratio of earnings before taxes and provisions

in the income statement to the value of total assets at the time of observation as the representative variable The purpose is to eliminate the effect of individual bank’s scale and size of operation A positive relationship between the variable and loan loss provisions will be consistent with the income smoothing hypothesis

2.2.1.2 Capital management

Capital management occurs when capital becomes constrained, banks will excise discretion on different accounting accounts, for example, loan write-offs, security gains, loan loss provisions, and equity to achieve regulatory capital targets Among those, loan loss provision account is the most popular capital management tool identified in literature

Valverde and Fernandez (2014) suggest that banks use loan loss provisions to alter their regulatory capital ratios in capital management The loan loss provisions have a positive effect on the regulatory capital level because it is part of the regulatory capital The capital management hypothesis projects that the regulatory capital ratio

is negatively associated with loan loss provisions because banks with low capital ratios can increase them by charging more loan loss provisions to decrease regulatory costs imposed by capital adequacy ratio regulations On the other hand, capital management may result in undesirable effects for bank risk management since it implies an artificial increase in capital ratios at the expense of a cutback in the coverage of expected losses By exerting capital management, banks decide the current loan loss provisions of the period and enable retained earnings to contribute

to reducing the distance between the objective and the level of regulatory capital

There have been conflicting results from empirical studies on bank capital management through loan loss provisions Moyer (1990) finds evidence that, before Basel I took effect, some bank managers adjusted the discretionary component of loan loss provisions to manipulate the capital adequacy ratio in order to reduce regulatory costs as well as avoid regulatory capital constraints Beatty et al (1995) contribute to the literature by developing a methodology that takes into account the

simultaneity of bank accounting, financing, and operating decisions They come to the conclusion that both loan loss provisions and loan charge-offs reflect not only loan quality evaluation but also capital management decisions Meanwhile, Ahmed

et al (1999) find that the closer banks are to breach the regulatory requirements, the higher the probability that they will engage in capital management Using loan growth

as a proxy for violating capital requirements, the authors show that banks with an above-median percentage change in loans in a given year manage capital via loan loss provisions in US banks The cross-country analysis by Bikker and Metzemakers (2005) suggests that capital management behavior exists in the US, Japan, and most

EU countries, although no evidence is found for Spain Perez et al (2006) also find

no support for capital management in Spanish banks The mixed results in testing the capital management hypothesis may be attributed to the fact that the amount of capital allocation in general provisions varies greatly across level of bank capitalization, countries, jurisdictions, and sample periods

In prior studies, capital adequacy ratio is commonly used to investigate the behavior

of capital management in banks However, in this thesis, as many commercial banks

in Vietnam provide limited information on the capital ratio, the author alternatively uses the ratio of equity to total assets to capture the capital management behavior A negative relationship between the variable and loan loss provisions will be in consistency with the capital management hypothesis

2.2.1.3 Signaling

Together with income smoothing, earnings management can be used as a signaling mechanism (Valverde and Fernandez, 2014) Bank managers may use loan loss provisions to manage earnings and signal private information about future prospects

If managers have information showing that the bank value is higher than the market value, such banks may use provisions as a signal of strength (the potential to absorb future losses) to revise the value upward Beaver et al (1989) propose that

an increase in loan loss provisions can indicate that "management perceives the earnings power of the bank to be sufficiently strong that it can withstand a ‘hit to earnings’ in the form of additional loan loss provisions" Their reasoning implies that

the increase in loan loss provisions conveys a signal of good news about the strength

of a bank's earnings Nonetheless, an increase in loan loss provisions can also be viewed as bad news because loan loss provisions serve as the primary source of information on loan default By controlling timely indicators of loan default such as change in nonperforming loans, loan loss allowance, and loan charge-offs, any excess loan loss provisions will include only the "good news" component (Wahlen, 1994; Liu and Ryan, 1995)

Bank managers have different propensities to exercise signaling because they face different conditions and have different incentives Kanagaretnam et al (2003) hypothesize that the propensity to signal varies negatively with bank size and positively with earnings variability, future investment opportunities, and information asymmetry Since variability of reported income is regarded as an indicator of risk, Barth (1994) argues that shareholders of banks will demand a higher risk premium for the increased riskiness perceived from a more variable income stream Therefore, bank managers with high earnings variability will have a higher motive to signal their future earnings prospects In addition, the information environment hypothesis suggests that larger banks are more carefully monitored by regulatory agencies and followed by more analysts and institutional investors Therefore, managers of larger banks will have less private information to signal through loan loss provisions Furthermore, information asymmetry and income volatility can have combined effect

on signaling decisions because earnings with higher fluctuation tends to worsen information asymmetry between the banks and outside investors Accordingly, if earnings volatility increases information asymmetry and thus results in undervaluation of the firm due to adverse selection, banks with high historical earnings variability have greater incentive to signal their information about their future favorable prospects

Several studies offer evidence that bank managers use provisions to signal their private information about banks’ future prospects For example, Beaver et al (1989) recognize that investors interpret an unexpected increase in loan loss provisions as a signal of a bank’s financial strength Liu and Ryan (1995) demonstrate that loan

portfolio composition is predictably associated with the market reaction to and anticipation of loan loss provisions More specifically, they find that the market reaction to loan loss provisions is positive for banks with a high proportion of large and frequently renegotiated loans and that the advance market projection of loan loss provisions is stronger for these banks More recently, Kanagaretnam et al (2001) find that managers of underrated banks use loan loss provisions to signal their banks’ future earnings prospects In contrast to the above findings, Ahmed et al (1999) find

no evidence of the signaling hypothesis Anandarajan et al (2007) use a sample of Australian banks and carry out a study on earnings management, capital management and signaling They also do not find significant use of loan loss provisions for signaling future purpose of earnings to investors Eng and Nabar (2007) find that unexpected loan loss provisions are positively related to bank stock returns and future cash flows, but this relationship tend to be significantly lower in the crisis years

In this thesis, following previous studies, one-year-ahead change of earnings before taxes and loan loss provisions is used to capture signaling behavior as in previous literature A positive relationship with loan loss provisions would indicate that banks might signal their future improvement of earnings to their clients and investors by increasing their provision amount

renegotiate with their borrowers than smaller banks do As a result, the larger the size

of a bank is, the more its loan loss provisions decrease In the study of Chang at el (2008), it is concluded that because large banks have more resources as well as abilities to make profits, and they are more likely to avoid using discretionary loan loss provisions to manipulate earnings On that account, the relationship between bank size and the use of loan loss provisions might be expected to be negative

In this thesis, the natural logarithm of total assets is used to test the potential influence

of bank size on loan loss provisions Either a positive or negative relationship between the variable and loan loss provisions may support the influence of bank size on banks’ provisions

In practice, loans are usually classified into different categories depending on the number of days overdue, following a regulatory framework For example, in Vietnam, Group 1 represents loans that are on schedule, Group 2 loans have 10 to 89 days overdue and Groups 3 to 5 represents nonperforming loans with 90 days to 1 year overdue Nonperforming loans are indicative of asset quality An analyst needs

to estimate the real default risk For example, unsecured consumer lending has higher default rates, but also has higher interest rates, meaning it may or may not be a profitable business line, depending on both factors One should expect different nonperforming loan ratios by segments, so the comparison of one bank to another requires taking into account the different client base Higher default can be acceptable

if it is compensated by higher interest rates charged

There have been several studies testing the influence of nonperforming loans on loan loss provisions with contrary results Collins et al (1995), Greenwalt and Sinkey

(1988), and Chang et al (2008) all find that there is a positive relationship between provisions and nonperforming loans In other words, as the amount of nonperforming loan increases, bank managers may increase the provisions for loan losses in order to reduce the ratio of nonperforming loans The result is in line with the findings of Taktak et al (2010), Bouvatier and Lepetit (2012), Valverde and Fernandez (2014), Adzis et al (2015), and Fernando and Ekanayake (2015) However, it contradicts the findings of Moyer (1990), Beatty et al (1995), Ahmed et al (1999), and Anandarajan

et al (2007)

In this thesis, nonperforming loans scaled in relation to total gross loans is used to test the influence of nonperforming loans on loan loss provisions A positive coefficient of the variable will be on agreement with the hypothesis that loan loss provisions are set up based on nonperforming loan amount

c) Loan growth

Loan growth is the change in the total amount of issued credits given by banks relative

to the prior accounting period and it is used as a proxy of the credit risk exposure of the bank in its intermediation activities (Bushman and Williams, 2012) As stated by Lobo and Yang (2001), the influence of this variable on loan loss provisions largely depends on the quality of incremental loans

Loan growth is considered as a proxy of the bank risk (Bikker and Metzemakers, 2005; Jimenez and Saurina, 2006) because the increase of the loans in economic upswings leads to an increase of the risk Banks pursuing higher lending growth rates are more likely to accept riskier borrowers Loans growth is associated with a fall of the efforts of banks’ monitoring and a deterioration of the quality of the portfolio Concurrently, the optimism of banks results in lower loan loss provisions than the amount should be As a result, a negative relationship between loan growth and loan loss provisions is suggested as in many prior studies (Cavallo and Majoni, 2001; Laeven and Majoni, 2003; and Dinamona, 2008)

On the other hand, a high rate of credit growth may also lead to the increase in loan loss provisions due to the increase in potential default risk imposed on banks Fonseca

and Gonzalez (2008), Bikker and Metzemakers (2009), and Dong et al (2012) all suggest positive coefficients of loan growth relative to loan loss provisions Besides,

a prudent bank that regularly makes careful adjustment of loan loss provisions according to the change in loan amount should show a positive association between the amount of loan loss provisions and the growth rate of its loan portfolio

In this thesis, loan growth is represented by the ratio of the difference between total loans of two consecutive years to average total assets of these years The purpose is also to eliminate the effect of individual bank’s scale and size of operation Either a positive or negative relationship between the variable and loan loss provisions will show that commercial banks in Vietnam may make decisions on provisions based on loan growth

d) Lag of loan loss provisions

Following Laeven and Majnoni (2003) and Fonseca and Gonzalez (2008), the author examines a lagged value of loan loss provisions, which is also the beginning loan loss provisions of the previous year , to capture the dynamic adjustment of loan loss provisions If banks adjust their provisions slowly to recognize potential losses against loans following a default event, then provisions could be systematically related to the previous period The lagged dependent variable takes into account a change in the speed of adjustment of loan loss provisions

Packer and Zhu (2012) find lag of loan loss provisions to be positively significant for banks in China, India, and South East Asia, suggesting “a certain degree of persistency in the time series of loan loss provisions” Japan is found to have an insignificant lagged coefficient, meaning a relationship between one period of provisions and the previous period does not exist Bikker and Metzemakers (2005) found that both Japanese banks and banks in Luxembourg did not dynamically provision either A positive coefficient implies that conventional banks do not make enough provisions at the start and slowly add to their provisions as they recognize losses following a default event

In this thesis, we consider a dynamic adjustment of loan loss provisions through the use of lagged independent variable derived from dependent variable If banks adjust their provisions slowly to recognize potential losses against loans or if default events are concentrated in time, then provisions could exhibit time dependency Therefore,

a positive relationship between the lagged variable and loan loss provisions is expected

2.2.2 Macroeconomic factors

Pro-cyclicality is the phenomenon in which financial conditions, which are usually expressed though GDP growth and consumer price index, tend to improve during economic upswings, amplifying the expansion, and then to deteriorate during downswings, intensifying the contraction (Fonseca and Gonzalez, 2008) Banks are considered to behave in a pro-cyclical manner since their actions tend to strengthen the momentum of underlying economic cycles – their lending and provisioning practices move in correlation with the short-term business cycles

During a cyclical upswing, banks tend to be overly optimistic about the economy and hence their customers’ position Banks advance loans against overrated collateral, reduce the applied risk premium and allocate less loan loss provisions to cover expected risks Meanwhile, there is usually a rise in bank profitability during an economic boom Subsequently, banks’ pro-cyclicality contributes to rapid credit growth, the rise in collateral values, artificially low lending spreads and a decline in loan loss provisions However, when business cycles trend down and the optimism exhibited during a cyclical upturn vanishes, formerly hidden problems become suddenly transparent At such times, banks will typically behave in a way that further worsens the situation – responding with an excessive cutback in lending, leading to a credit crunch, or setting up disproportionately large loan loss provisions, which can undermine their profitability and aggravate their capital situation In extreme scenarios, banks’ pro-cyclical behavior can even precipitate a system-wide banking crisis

In practice, the level of provisioning has a historically pro-cyclical bias As it is basically linked to contemporaneous problem assets, provisions mainly rise during a

downturn when credit risk has already materialized There are some factors that contribute to its pro-cyclicality In one sense, business cycle developments are hard

to identify, and therefore there may exist a disconnection between the timing of loan loss provisioning and the assessment of credit risk On the other hand, accounting frameworks only allow provisioning for losses that have already been incurred as of

a financial statement date, which does not really address the concept of “expected losses” Perez et al (2006) note that general provisions usually rise during an economic boom when banks give out more loans and the demand for credit is high during this period During a downturn, loans to riskier companies would incur larger loan losses because risks materialize, and therefore higher specific loan loss provisions follow

Several empirical studies have been carried out to address the issue of pro-cyclicality

of loan loss provisioning of the banking system Bikker and Hu (2002) study the cyclicality behavior of OECD banks under the Basel I regime by investigating the effect of business cycle on bank profits, loan loss provisioning, and bank lending of OECD countries, including Australia and New Zealand The findings indicate that loan loss provisions are high during economic downturns, which support the evidence

pro-of pro-cyclicality, but they lessen if the bank’s net income is relatively high The same result is pointed out in the studies of Laeven and Majoni (2003), Bikker and Metzemakers (2005) The findings of Handorf and Zhu (2006), however, do not support the evidence of pro-cyclical behavior of loan loss provisioning for average-sized banks, which may also indicate that these banks engage in income smoothing activities that lead to the countercyclical pattern Nonetheless, the existence of pro-cyclicality is found among smaller banks and the largest banks in the US Bouvatier and Lepetit (2008) find that nondiscretionary component of loan loss provisions has

a significant relationship with the business cycle They also strongly support the implementation of dynamic provisioning as is applied in Spain as it promotes a forward-looking approach and hence reduces the pro-cyclical behavior of loan loss provisions

In this thesis, as in other studies, GDP growth rate and inflation rate are the two independent variables used to test the pro-cyclicality behavior through loan loss provisions of Vietnamese commercial banks Since economic upswings are usually indicated by the increase in GDP and inflation, these two variables are expected to have negative association with loan loss provisions Besides, because the time scope

of the thesis covers the period within which the global financial crisis 2007-2009 took place, the dummy global financial crisis is added to the regression model Therefore, there are three macro-economic explanatory variables in total

In summary, in this chapter, we have had an insight into loan loss provision and its expected determinants The definition, related literature, and their representing variables, either dependent or independent, have been meticulously reviewed based

on theory established by researches mostly over the last three decades These researches offer an eclectic wide range of observations regarding loan loss provisions, which alter in both spaces, times, and perspectives In this chapter, the expected relationships between dependent and independent variables are also summarized Loan loss provision is expected to be positively influenced by income smoothing, signaling, and other bank-specific factors such as nonperforming loans and one-year lag of loan loss provision Meanwhile, it is predicted to be negatively associated with capital management and pro-cyclicality The remaining bank-specific factors bank size and loan growth are not strongly proved to be positively or negatively related to loan loss provision in prior literature Therefore, both positive and negative signs may

be expected for these variables

Chapter 3: METHODOLOGY

3.1 Data and sample

The thesis uses secondary bank accounting data extracted from consolidated income statements and balance sheets of selected banks The banks’ financial information was obtained primarily from bank annual and financial reports published on the website vietstock.vn The data were carefully reviewed to avoid double counting of institutions, to ensure that the banks reported in accordance with the same accounting standards Macroeconomic data such as GDP growth rate and inflation rate were taken from World Bank database A few banks with less than three consecutive time series observations had to be excluded from the sample in order to explore in a robust way the phenomena from not only a cross-sectional but also a dynamic point of view The final data set consists of annual end-of-year information for all domestic commercial banks covering the period 2007 to 2014 The eight-year period of 2007-

2014 is efficiently long enough to capture both an economic upswing and a downturn This specific time interval captures the downturn caused by the financial crisis from

2007 to 2009 and the following recovery from crisis The final sample comprised a total of 39 commercial banks with 312 bank-year observations

This thesis utilizes an unbalanced panel data set This refers to a sample in which some cross-sectional units (in this study, it refers to banks) have an unequal number

of time-series observations Such incomplete panels are common in an economic empirical setting and particularly in banking industry, incomplete panels are unavoidable because some banks have to be dropped out from the sample due to bankruptcy, mergers and acquisitions

Panel data provides several advantages Firstly, it controls for bank- and time- invariant variables Secondly, it offers richer data with more variability and less collinearity among the variables Thirdly, it gives greater freedom and efficiency Fourthly, it enables the use of a dynamic model Fifthly, it identifies and measures effects that cannot be detected in cross-section and time-series data Sixthly, it tests more complicated behavioral models Finally, it reduces biases resulting from aggregation over firms or individual The only limitations are that panel data may

cause measurement error problems and may exhibit bias due to sample selection issues (Adzis, 2012)

3.2 Hypothesis development

The empirical evidence from earlier international studies indicates that banks generally use loan loss provisions for income smoothing, capital management, and signaling These manipulating activities happen because bank managers know more about what happens inside the bank than outsiders do and they have substantial discretion on loan loss provisioning Hence the hypotheses regarding income smoothing, capital management, and signaling are developed as follows:

H1: Vietnamese commercial banks smooth income via loan loss provisions

H2: Vietnamese commercial banks manage regulatory capital via loan loss provisions

H3: Vietnamese commercial banks engage in signaling via loan loss provisions

A range of hypotheses for the relationship between loan loss provisions of Vietnamese commercial banks and the other bank-specific factors has also been set up:

H4: Loan loss provision is positively/negatively related to bank size

H5: Loan loss provision is positively related to nonperforming loan

H6: Loan loss provision is positively/negatively related to loan growth

H7: Loan loss provisions is positively related to loan loss provisions of the previous period

From the aforementioned studies, it can be concluded that loan loss provisions is influenced by business cycle, which is considered as pro-cyclicality Thus, the final hypothesis is developed as follows:

H8: Vietnamese commercial banks exhibit pro-cyclical behavior via loan loss provisions

3.3 Variables