( PDF ) Sale of credit portfolio and risk: the case of financial institutions in Brazil

Download PDF

ads:

FUNDAÇÃO GETULIO VARGAS

ESCOLA DE ADMINISTRAÇÃO DE EMPRESAS DE SÃO PAULO

GRAZIELLA LAGE LAUREANO

SALE OF CREDIT PORTFOLIO AND RISK: THE CASE OF FINANCIAL

INSTITUTIONS IN BRAZIL

SÃO PAULO

2009

ads:

Livros Grátis

http://www.livrosgratis.com.br

Milhares de livros grátis para download.

GRAZIELLA LAGE LAUREANO

SALE OF CREDIT PORTFOLIO AND RISK: THE CASE OF FINANCIAL

INSTITUTIONS IN BRAZIL

Dissertação apresentada à Escola de

Administração de Empresas de São Paulo da

Fundação Getúlio Vargas, como requisito para

a obtenção de título de Mestre em

Administração de Empresas

Campo de Conhecimento:

Mercados Financeiros e Finanças Corporativas

Orientador:

Prof. Dr. Richard Saito

SÃO PAULO

2009

ads:

Laureano, Graziella.

SALE OF CREDIT PORTFOLIO AND RISK: THE CASE OF FINANCIAL

INSTITUTIONS IN BRAZIL / Graziella Lage Laureano. - 2009

42 f.

Orientador: Richard Saito.

Dissertação (mestrado) - Escola de Administração de Empresas de São

Paulo.

1. Título de créditos -- Brasil. 2. Instituições financeiras -- Administração.

3. Securitização. 4. Risco (Economia). I. Saito, Richard. II. Dissertação

(mestrado) - Escola de Administração de Empresas de São Paulo. III. Título.

CDU 336.763(81)

GRAZIELLA LAGE LAUREANO

SALE OF CREDIT PORTFOLIO AND RISK: THE CASE OF FINANCIAL

INSTITUTIONS IN BRAZIL

Dissertação apresentada à Escola de

Administração de Empresas de São Paulo da

Fundação Getúlio Vargas, como requisito para

obtenção de título de Mestre em Administração

de Empresas

Campo de Conhecimento:

Mercados Financeiros e Finanças Corporativas

Orientador: Prof. Dr. Richard Saito

Data da Aprovação:

_____ / _____/ ________

Banca Examinadora:

Prof. Dr. Richard Saito (Orientador)

FGV – EAESP

Prof. Dr. Rodrigo De Losso

FGV – EAESP

Prof. Dr. Roberto Calfat

IBMEC - SP

AGRADECIMENTOS

Agradeço ao Professor Doutor Richard Saito pela orientação e ensinamentos

transmitidos.

Agradeço aos professores Rafael Schiozer e Verônica Fernandez por toda a ajuda

que deram ao longo do desenvolvimento desse trabalho.

Agradeço aos professores Rodrigo De Losso, Afonso Pinto e novamente ao

professor Rafael Schiozer por suas contribuições na ocasião da defesa da proposta.

Agradeço à Débora e ao Uverlan pelas ajudas com a estruturação da base de

dados.

Agradeço a Deus por ter me guiado até aqui.

Agradeço à minha família pelo amor e o apoio de sempre.

Agradeço aos meus amigos pela compreensão e ajuda, em especial àqueles que de

alguma forma colaboraram para a realização deste trabalho.

Agradeço aos professores Aureliano Bressan e Renato Assunção pelo incentivo ao

ingresso no mestrado.

Agradeço à CAPES pela ajuda financeira concedida durante o curso.

Agradeço ao Centro de Excelência Bancária - FGV pelo apoio ao projeto e auxílio

financeiro.

RESUMO

Este estudo analisa se as vendas de carteiras de crédito são utilizadas por

instituições financeiras para gestão de risco, de acordo com Stanton(1998) e

Murray(2001) ou para captação recursos, como apontado em Cebenoyan e

Strahan(2001) e Dionne e Harchaoui(2003). Duas hipóteses foram testadas quanto

às vendas de carteira de crédito: 1) implicam em melhor rating na carteira

remanescente; ou 2) promovem alavancagem financeira - com piora na carteira

remanescente -, controlando para a existência de coobrigação e para quem esses

ativos foram transferidos. A amostra inclui informações trimestrais de 145

instituições financeiras do primeiro trimestre de 2001 ao segundo trimestre de 2008.

Os resultados oferecem evidências empíricas de que as instituições financeiras

utilizam estas vendas para melhora do rating da carteira de crédito remanescente,

ou seja, elas transferem, em sua maioria, ativos de baixa qualidade, garantindo bons

ratings e melhorando a liquidez.

Adicionalmente, seguindo a proposta Dionne e Harchaoui(2003) - que além de

testar, evidenciam que exigências regulatórias promovem alavancagem em ativos

de alto risco - foi observada a relação entre o Índice de Basiléia e rating da carteira

de crédito. As conclusões foram semelhantes às encontradas por Dionne e

Harchaoui(2003): quanto mais adequada – maior Índice de Basiléia - uma instituição

financeira for, maiores as chances de ela possuir uma carteira de crédito com

qualidade ruim.

Palavras-chave: venda de carteira de crédito, recebíveis, securitização,

alavancagem e mitigação de risco.

ABSTRACT

This study examines whether the sale of credit portfolios are used by financial

institutions for risk management, according to Stanton (1998) and Murray (2001) or

to capture resources, as indicated in Cebenoyan and Strahan (2001) and Dionne and

Harchaoui (2003). Two hypotheses on credit portfolio sales were tested: 1) promote

rating improvement to the remaining portfolio, or 2) drive to financial leverage - with

worsening on the remaining portfolio -, controlling for recourse existence and to

whom those assets were sold. The sample includes quarterly information from 145

financial institutions from the first quarter 2001 to second quarter of 2008. The results

provide empirical evidence that financial institutions use these sales to improve the

rating of the remaining credit portfolio, i.e. they transfer, in most cases, low quality

assets, assuring good ratings and improving liquidity.

Additionally, following Harchaoui and Dionne (2003) proposal - which besides testing,

demonstrating that regulatory requirements promote leveraging in high-risk assets –

it was observed the relationship between the Basel Index and credit portfolio rating.

The conclusions were similar to those found by Dionne and Harchaoui(2003): the

more adequate – higher Basel Index - the financial institution, the greater its chances

of having a bad quality credit portfolio.

Keywords: credit portfolio sales, receivables, securitization, leveraging and risk

mitigation.

INDEX OF TABLES

Table I – Hypotheses and Expected Signs..………………………..……..…………….19

Table II – Credit Rating………………………………………………..….………………..22

Table III – Statistics of Sales of credit Portfolios…………………………. ...………….28

Table IV – Sales of Credit Portfolio, Rating and Capital Adequacy…...……..……….29

Table V – Sales of Credit Portfolio and Remaining Portfolio Structure ……………...31

Table VI – Summary of Empirical Evidence on Changes in Portfolio…..…….....…...34

Table VII – Sales of Credit Portfolio and Total Remaining Portfolios……….…..........35

Table VIII – Sales of Credit Portfolio and Profitability……………….....….……………38

INDEX

INTRODUCTION................................................................................................. 9

AN OVERVIEW OF SALES OF CREDIT PORTFOLIO.................................... 10

2.1

Securitization Workflow............................................................................................10

2.2

Receivable Sales Advantages ....................................................................................12

LITERATURE REVIEW..................................................................................... 14

HYPOTHESES AND METHODOLOGY............................................................ 16

4.1

Do credit portfolio sales make the remaining portfolio rating of the Brazilian

financial institutions better or worse?...................................................................................17

4.2

Do credit portfolio sales promote changes in loans volumes? If so, what kind of

loans: The best or worst loans portfolio? .............................................................................18

4.3

How do credit portfolio sales affect ROE?................................................................19

4.4

How does capital adequacy affect credit portfolio ratings?.......................................20

4.5

Sample and Panel Models..........................................................................................21

RESULTS.......................................................................................................... 30

5.1

Rating and Capital Adequacy....................................................................................30

5.2

Loans Volume Change...............................................................................................31

5.3

Profitability................................................................................................................37

CONCLUSIONS................................................................................................ 39

REFERENCES.................................................................................................. 41

1 INTRODUCTION

When in need of liquidity, many financial institutions may either look for funding by

conventional ways or sell assets, like credit portfolios. By selling receivables to third

parties, the banks may mitigate or increase risk, depending on how those funds are

used. This study intends to analyze this behavior, considering a specific kind of

receivable sale will be considered: the sale of credit portfolio.

Common to worldwide practice, financial institutions use securitization to convert

a relatively homogeneous credit portfolio into securities. It is important to clearly

distinguish receivables sale and securitization: the first is a traditional selling practice

of receivables, while the second involves a transformation of these receivables credit

quality, enhancing their liquidity, reducing their credit risk and restructuring their cash

flows.

There are five reasons for credit portfolio sales: risk management; funding;

reduction of the average cost of capital; advantages over capital regulation; and

improved profits. These reasons are sort of advantages and solutions for banks

difficulties, what make sales of credit portfolio very attractive.

Thus, studies that examine the ideas behind and the purpose of credit portfolios

and securitization are becoming necessary for the world and Brazilian financial

markets. They could provide information, helping financial institutions on decisions

about selling portfolios and investors on decisions about buying financial institution’s

shares, facing that some practices imply on profitability improvement and/or risk

management, practices that add value to these companies. As an example of

financial institutions and investors’ benefits, assume that receivable sales always

imply on profitability gains but also increase risk: depending on how risky and

profitable a company is, an investor may think it is or not a good practice deciding to

buy or sell its stocks. Looking at financial institutions’ benefits, the riskier companies

would avoid doing receivable sales, while the less risky ones would do it as much as

possible. Instead, if receivable sales are always done to mitigate risk, riskier

companies would take more advantages from doing it than less risky ones.

Also, adding regulatory variables to such studies, supervisory bodies could watch

financial institutions’ behaviors and adequate their policies to completely reach their

regulation objectives. As receivable sales are operations that transfer risks, there is a

capital requirement reduction that allows financial institutions to sell receivables and

remove them from their balance sheets. This capital advantage is an incentive of risk

mitigation and brings many benefits to the companies that use it, but one of them

may not be desired by the supervisory bodies: the usage of the resources earned by

the sales on operations that increase institutions’ risks. If this is being heavily done,

the incentive stops reaching its purposes and needs to be reviewed or even

suspended.

Given the context and the reasons for credit portfolio sales, this study tested if

they are performed with the objective of providing funding or if they are done to

mitigate risks by Brazilian financial institutions . It was shown that they are mostly

done to mitigate risks instead of leveraging, but that there are some leveraging

operations that are always done with recourse. About profitability, it was found that

the leveraging types of portfolio sales promote it, reinforcing that leveraging decisions

are done on the purpose of increasing profits. Parallel to this, there is another

remarkable result where the most adequate financial institutions in terms of rating are

the ones that have the worst credit portfolios at their balance sheets, Basel Index,

what points that such capital requirement helps on illiquidity risk mitigation, but

pushes a bank to be credit riskier.

Therefore, this article is structured as follows: an overview of credit portfolio sales

in chapter two, literature review in chapter three, hypotheses and methodological

procedures in chapter four, empirical results in chapter five and conclusions in

chapter six.

2 AN OVERVIEW OF SALES OF CREDIT PORTFOLIO

2.1 Securitization Workflow

As a way of changing relatively illiquid assets into liquid assets, financial

institutions in the U.S. transfer risks by pooling loans to investors, selling them as

bonds. The bonds from credit portfolio sales are therefore characterized by a

commitment to future payment of a principal and interest, ballasted by the credit

portfolio and the expected cash flow that will come from the sale. Thus, to assign

loans means to convert some rights to ballast for securities issuing.

A typical future flow structure is set out in Chart 1 and involves:

• Product: Loans - Ballast for future receivable operation.

• The originating entity sells credit portfolio directly or indirectly to an Special

Purpose Vehicle – SPV – that promotes the securitization.

• The SPV issues the debt instrument and sells them to investors.

• Customers are directed to pay for the product from the originating entity

directly to the SPV.

• Excess collections from obligors are directed to the originator via SPV.

Chart I: General Securitization Process

$$$

Product

(Future

Receivable)

Rec. Sales

Debt

Instruments

$$$ $$$

In Brazil, bonds are not commonly issued and receivables sales are pooled by

Special Purpose Vehicles (SPV) or sent to funds called FIDCs – that actually are a

SPV type - and are not, therefore, directly negotiated in capital markets. Catão

Rodrigues and Libonati(2008) presented the following detailed workflow for the

Brazilian Structure involving FIDCS:

• The Financial institution provides loans to customers, transferring resources

and creating a credit portfolio (receivables).

• The rating agency analyzes and classifies the portfolio.

• The Fund issues senior and subordinated debts instruments (subordinated

ones are a sort of credit enhancement against portfolio defaults). The

subordinated shares are subscribed by the own financial institution while the

seniors are sold to investors. With these resources, the fund pays the

receivables to the transferor financial institution.

SPV

Investor

Originator

Customer

• The financial institution’s customers perform the payment of their loans, which

are deposited in a trustee financial institution account, responsible for

controlling the flow of financial transaction.

• The trustee, by a pre-established scheduled, pays firstly the investors. The

remaining money goes to the fund to cover its costs.

2.2 Receivable Sales Advantages

• Risk Management

By selling credit portfolio, financial institutions transfer to investors their inherent

risks, such as credit, liquidity and market risks, practicing portfolio risk mitigation.

The credit risk would be the borrower default chance or, in other words, the

probability of the financial institution not receiving back, form the costumer, the cash

lent on the loan operation.

The liquidity risk would happen in case of assets and liabilities mismatch, what

would affect the institution liabilities payment capacity. For the case of credit

portfolios without securitization, the cash flow consists of the net between its credit

portfolio its loans taken, not being always possible to balance them in a way to have

the smallest fluctuation. Securitizing their credit portfolio, their cash flows are

consequently transferred to the investor.

The market risk is the probability of occurring financial market conditions change

that would affect interest ratings and/or currencies’ changes and, consequently, the

amount of credit portfolio. For example, if for a loan it was agreed a fixed amount in

the future but the fund to finance it is ballasted in floating amounts, a change on the

interest rating will affect the assets and liabilities differently and may implicate on the

institution distress. Again, securitizing the credit portfolio, their cash flows are

transferred with its risk to the investor.

• Funding and Capital Structure

Assets sold in a securitization are exchanged for money, which will be used as

extra capital to finance new operations, implying in a reduction of debt demand.

Cebenoyan and Strahan(2001) gave an important contribution about credit portfolio

sales and capital structure by detecting that banks engaged on securitizations hold

less capital than the ones that do not do it.

• Reduction of Cost of Capital

If the spread charged by SPV to buy credit portfolio from financial institutions are

lower than the financial institutions cost of capital, this also reduces their final

average cost of capital. Furthermore, according to Fabozzi et al. (2006),

securitization is stratinggic way of accessing the capital market, what allows the

institutions to be known and facilitating them to, further, be stock issuers, what would

reduce even more their costs of capital.

• Advantages over Capital Regulation

Credit portfolio sales can be done in two basic ways: with or without recourse.

The difference is that the transferor may or may not take on part of the sold portfolio

risk. Securato (2002) defines sales without recourse as a permanent sale of assets

where the transferee has no rights over the debt sold. On sales with recourse the

Special Purpose Vehicle has recourse rights over the originators in cases of credit

default, a situation in which the SPV may directly contact the borrower or the financial

institution originator. When the portfolio is sold with recourse, the originator is

required to make provisions on the balance sheet; but when the sale is without

recourse, the portfolio is not required to be recorded and the originator may take

advantages over capital regulation. These advantages are taken when, by removing

loans from their balance sheets, there is a reflection of lowered risk and therefore the

amount of capital financial institutions must hold, given the size of their loans

portfolio, becomes smaller. According to Pinheiro (2008), such operations allow small

institutions to have a high volume of credit operation.

• Improved Profits

According to Pinheiro(2008), securitization permits a company to originate more

operations keeping the same level of capitalization, making its return on equity

higher. Also, a manager can choose to make the portfolio’s expected profit higher. To

do this, he would keep the high performance loans – usually the riskier ones - and

sell the low performance ones.

A benefit that comes together with profit improvement is that investors positively

valuate the gain of profitability and the stock price also increases.

3 LITERATURE REVIEW

Securitization bibliography usually analyzes securitization purposes and its effects

in the market. Some of them consider extra conditions such as crisis, arbitrage,

asymmetric information and regulatory procedures. They are very diverse and do not

follow the same theories, being mostly consisted of empirical analyses. So, this

literature review will be about the bibliographies focused on this study’s hypotheses:

risk mitigation x risk increase. Also it will be analyzed references for capital adequacy

and portfolio quality.

For risk mitigation, Stanton(1998), doing a study with commercial banks from

1985 to 1992 found that securitization is a way of alleviating the problem of

underinvestment since the sale of assets may be used to reduce the differences

between loans linked to projects with negative and positive correlation with the bank

existing assets payoffs. Murray(2001) performing a critical bibliographic review,

observing if securitizations promote excessive credit creation, asset monitoring,

liquidity illusion, monetary policy and risk transference illusion, finds that they do not

increased risk and is a source of stability rather than instability for the financial

system. Tomas and Wang(2004) using the BHC database from the Federal Reserve

Bank of Chicago for the period between 1992 to 2000 investigated why banks cede

assets in order to manage risks, applying a model with different credit portfolio sales

types in periods of liquidity shock and credit shock. They found that credit portfolio

sales with and without risk retention are used respectively to avoid liquidity and

capital shocks. Unfortunately, there was no evidence that they help on credit shocks,

but an important point for this study was that having or not risk retention determines

different usages and objectives.

For the leveraging hypothesis, Cebenoyan and Strahan(2001) test how access to

the market of the securitization affects the decisions about capital structure and credit

policies of the U.S. domestic commercial banks from June 1987 to the end of 1993.

They try to detect whether banks that are more able to work on the credit risks in the

securitization market get better benefits. As a result, they found that in fact those

banks do get better benefits: in particular, banks that sell and purchase loans have

less capital, higher leverage and lend more to high-risk borrowers than the ones that

only sell or even do not perform such operations. Dionne and Harchaoui(2003)

investigate the relationship between banks’ capital, securitization and risk for

Canadian banks between 1988 and 1998 in the context of rapid growth of balance-

sheet activities, showing that securitization has a positive relation to banks’ risks.

Dionne and Harchaoui(2003) also is a reference for capital adequacy. They show

that securitization has a negative relation with two types of capital adequacy ratings

that constitute the capital adequacy Basel ratio what leads to the conclusion that

banks might be induced to shift to more risky assets under current capital

requirements. Minton, Sanders and Strahan(2001), also check between other tests, if

regulatory distortions in the Basel Capital Accord create incentives for securitization,

specifically for highly levered banks, in order to avoid binding capital requirements.

They find that unregulated finance companies and investment banks are much more

apt to securitize assets than banks, and that risky and highly levered financial

institutions are more likely to engage in securitization than safer ones. At the same

time, highly leveraged banks are less likely than better capitalized banks to

securitize. Bannier and Hänsel (2007) studied the determinants of securitization,

focusing on collateralized loan obligation of European banks’ from 1997 to 2004 in

order to identify macro-economic and firm specific factors. Between others, one of

their conclusions was that the activity of regulation exercised by the Basel I does not

push the market to securitization. Thus, contrarily from Dionne and Harchaoui (2003),

the two other studies present evidence pointing that securitization does not

compensate poorly structured regulations.

In Brazil, Pinheiro (2008) analyses, by Monte-Carlo simulations, interest rating

and default risks associated to funds’ senior and subordinated quotes, concluding

that, for the senior quotes, both risks are remote and for subordinate ones they are

very low. Catão, Rodrigues and Libonati (2008) verified the existence of a

relationship between credit portfolio securitizations and leverage, liquidity and quality

of the portfolio credit ratios. They utilized some data from the Central Bank, same

information feeder that this study utilizes, but considered only a sample of 10

Brazilian banks, analyzing case by case along 18 quarters. Their results revealed

that 7 out of 10 presented significant relations, pointing out that securitization

influences those issues.

4 HYPOTHESES AND METHODOLOGY

This study objective is to identify if credit portfolios sales are mostly used as a

way of managing or increasing risk. To do this, some tests will be performed to

determine if these sales lead changes to the credit portfolio rating, on operations’

volumes and financial institutions’ profitability.

Cebenoyan and Strahan(2001) found the US domestic commercial banks that the

deals credit sales have higher leverage and lend more to high risk borrowers than the

ones that do not do it. Contrarily to Staton(1998) and Murray(2001), that suggested

that banks use receivable sales to reduce differences between loan types and

promote stability - being therefore a risk management instrument - they suggest that

banks use receivable sales as a way of leveraging, and consequently increase their

risks.

Given those contrary hypotheses observed in the literature, the empirical tests of

this study are expected to detect a tendency for different kinds of credit portfolios

sale, supporting, for each of them, one of the following two approaches:

I) Credit portfolio sale is a way for risk mitigation and its practice reduces portfolio

risk;

II) Once ceding, financial institutions get the chance to increase their loan

volumes and accept a bigger share of high-risk borrowers, promoting an increase in

credit portfolio risk.

Parallel to this, the returns of financial institutions that use credit portfolio sales

will be examined to reinforce those results, checking if the credit portfolio sales that

promote leveraging also promote profitability, as this is a profitable way of financing

credit portfolios.

In summary, this study attempts to detect risk management and risk increasing

behavior by running some tests to answer the following four questions:

• Do sales of credit portfolio make the remaining portfolio rating of the

Brazilian financial institutions better or worse?

• Do sales of credit portfoliopromote an increase in loans volumes? If so, on

what kind of loans: The best or worst loans portfolio?

• How do sales of credit portfolio affect ROE?

It is also this studies intent to observe the relation between capital adequacy and

the credit portfolio rating and try to detect if capital adequacy have influence over the

credit portfolio’s structure. To check this, a Basel regulation Index called CAR -

Capital Adequacy Ratio - built to push financial institutions to have enough liquid

assets, given their assets risks, will be used. For this case, the question to be

answered is:

• How does capital adequacy affect credit portfolio rating?

The justification of each question will be done, one by one, on the following

sections:

4.1 Do credit portfolio sales make the remaining portfolio rating of the

Brazilian financial institutions better or worse?

Usually, whenever a financial institution decides to cede its loans, it is necessary

to decide what kinds of loans, in terms of rating, will be sold.

The ratings classify the quality of the portfolio, or its level of risk, which depends

on the chances of default. In Brazil, credit portfolio ratings go from AA to H. For

example, if a portfolio is very good and borrowers are the best payers, its rating will

be AA; otherwise, as the chances of paying decrease, the classification turns to A, B,

C and so on. These ratings depend on several issues and one of them is the delay

for credit payments. Ratings AA and A have one similar issue: they are portfolios

without delays.

High rated portfolios, when sold, are better paid than low rated portfolios. That is

why financial institutions would prefer to cede the higher rated ones.

So the main idea here is to understand if credit portfolio sales make the rating of

the total remaining credit portfolio better or worse and consequently detect whether

portfolio sales are being used to mitigate risk by improving the remaining portfolio

quality or to earn resources by making it lower. Therefore, if the hypothesis of risk

mitigation is true, a positive relation between credit portfolio sales and the toal

remaining portfolio rating of the remaining portfolio is expected; if the hypothesis of

risk addition is true, a negative relation is expected. In other words, a positive relation

between credit portfolio sales and portfolio rating could point out that financial

institutions are probably securitizing bad rated portfolios. On the other hand, if a good

payer loan is sold the rating of the remaining portfolio is likely to get worse.

The use of different groups of credit portfolio sales will allow a more detailed

study that will check if sales of good assets and bad assets are related to specific

sale groups.

4.2 Do credit portfolio sales promote changes in loans volumes? If so, what

kind of loans: The best or worst loans portfolio?

It is important to clarify that the data utilized distinguishes sales by SPVs and

recourse, but does not reveal the sold portfolio exactly consistency. To solve this

difficulty, generic volumes of portfolio sales are compared to changes occurred on

the remaining specific credit portfolios, using the relative change of loans’ volumes -

according to its quality classification and the total change of loans’ volume - as the

explained variable.

Selling loans at first promotes a reduction of the remaining credit portfolio.

Similarly, a sale of a specific kind of portfolio also reduces its amounts and its relative

amount over the whole portfolio. This reveals the connection between credit portfolio

sales and portfolios’ growth: if the relation between credit portfolio sales and one type

of portfolio growth is negative, it means those sold loans were of this same type. A

positive relation shows that the sale increased that type of portfolio participation over

the whole portfolio which means that another kind, but not that specific one, was

sold. This second relation does not tell much about selling loans of that specific

portfolio.

To check the credit sales effects through hypotheses - risk mitigation against risk

increase - it is important to be clear that the expected signals – negative or positive -

for the types of credit portfolio are different. If the hypothesis of risk mitigation is true,

financial institutions should sell the worst credit portfolios. In this case, a negative

relation between credit portfolio sales and the worst portfolios would be expected.

Instead, if the hypothesis of cheap funding and risk addition is true, a negative

relation between credit portfolio sales and the best credit portfolios is expected.

Additionally, the credit portfolio sales will be subdivided in groups of linked or non-

linked, securitizer or financial institution, and with or without recourse, totaling eight

different groups. With those groups the analysis will not only be about credit portfolio

sales but the credit portfolio sales types. For example: for a change on the best

portfolio loans volume, the recourse sales to linked securitizers influence could be

negative while the without recourse sales to non linked financial institutions influence

could be positive.

4.3 How do credit portfolio sales affect ROE?

This question will be analyzed to reinforce the preview questions’ answers. If the

hypothesis of risk mitigation is true, financial institutions should not use the sources

earned by the credit portfolio sales to deal new riskier operations and, therefore,

should not have their profitability increased by it and no significant or negative effect

on the results is expected. The negative relation is justified because when a financial

institution cedes the right to receive future cash flows, accepting an amount now for

it, it may be giving away part of the operation’s profitability.

Instead, if the hypothesis of funding and risk addition is true, a positive relation is

expected as capital capitation from selling loans allows new profitable operations.

The next three questions on the table below expose hypotheses and behaviors

that would point out if the financial institutions are practicing credit portfolio sales with

the main objective of mitigating risk or leveraging with brief comments.

Table I: Hypotheses

Hypotheses will be empirically tested through panel data regression models for the period between January 2001 and

September 2008. Independent variables – described in the Hypothesis column – have their expected signs according to the

second column for risk management and to the third column for risk increase. Brief comments are on the fourth column.

Hypotheses Expected Signs

Risk Manag. Risk Incr.

Comments

Stanton( 1986)

and Murray(2001)

Dionne and

Harchaoui(2003)

Portfolio Rating

improvement

If credit portfolio sales are done to mitigate

risk their remaining Portfolio Rating will get

better; if they are done to leverage, what

increases risk, it would get worse.

Change on loans

volume

Better loans

Negative relation between change on good

credit portfolios and credit portfolio sales

means that this kind of portfolio is being sold.

Worse loans

Negative relation between change on bad

credit portfolios and credit portfolio sales

means that this kind of portfolio is being sold.

Profitability

improvement

If credit portfolio sales are done with leverage

aims, there will be an improvement on

profitability, if not there will be a reduction on

it.

4.4 How does capital adequacy affect credit portfolio ratings?

Capital adequacy ratio (CAR) is a regulatory ratio where, according to the Brazil

Central Bank, financial institutions have to hold capital against liquidity risk and have

Basel criteria sets of at least 11%. So the higher CAR is, the better financial

institution liquidity will be.

Minton, Sanders and Strahan(2004) and Bannier and Hansel(2007) analyzes a

hypotheses of securitization incentives created by the Basel Capital Agreement in

order to avoid binding capital requirements.

Dionne and Harchaoui(2003) gave a contrary contribution identifying that banks

under capital requirements have the riskiest loans portfolios.

Another hypothesis analyzed in this study is about capital adequacy and risk

mitigation: This is done by testing if, once capital adequacy, a type of Risk Mitigation,

is being done, financial institutions lend to worse payers to compensate the loss of

profit on capital withheld to keep CAR’s level. As a liquidity warranty, a CAR over

11% is a way to manage liquidity risks. Keeping the credit portfolio rating high and,

therefore, avoiding a default crisis is another. By managing risks with both

techniques, financial institutions would be better protected but they would also be

giving up more profitability than required by legislation. Once financial institutions

achieve the 11% ratio, thus giving up the chance to get high interest back from

investing in other riskier asset allocations, they will probably try to earn money by

other ways and the low rating operations could be one of them.

So, due to the need of capital to feet capital adequacy and the risk-profit trade-off

in the context of liquidity regulation, there is an expected negative relation between

the credit portfolio rating and CAR.

4.5 Sample and Panel Models

All the data used are publicly available at the Brazilian Central Bank website. The

Data base has quarterly information about Brazilian financial institutions from the first

quarter of 2001 to the second quarter of 2008. In summary, the original data-base

has 30 quarters and 145 financial institutions, totaling 4.350 financial institutions-

quarters.

To represent credit portfolio sales quarterly credit portfolio sales will be used,

discriminated by groups. This variable, CAR, and interest rating, are not pieces of

information that come from balance sheets, while all remaining variables –

dependent and of control - have their calculus based on preview information

extracted from them.

It is important to highlight that, for this study, many different types of data, to build

an unpublished and a huge Brazilian financial institution database, were assembled.

In light of the fact that there are not many studies about credit portfolio sales and risk

in Brazil, using such a rich database, this study demonstratings its relevance.

To examine the influence of credit portfolio sales over financial institutions’ risks,

credit portfolio sales volume will be regressed, along with control variables, against

the following dependent variables: rating, four groups of loans volume change, total

loans volume change and profitability.

As we also want to test if capital adequacy does or does not make the remaining

portfolio rating worse, the Capital Adequacy Ratio – CAR - will be one of the

independent variables regressed against the rating of the whole portfolio.

The models need to be controlled for endogeneity. Endogeneity is a modeling

problem that happens when the dependent variable may explain the independent

one. In this case, the rating of the portfolio, the loans’ volume change and profitability

are factors that may have influence over the decision of selling loans.

As the objectives of the models are to find the real influence of credit portfolio

sales over these variables, the endogeneity problem needs to be treated to allow this

influence to be correctly detected. To do this, Arellano and Bond was used. This is a

technique for panel data that uses an entire set of lagged variables as instruments,

providing exogeneity to the models. Arellano and Bond models use instruments to

represent the lagged first difference of the dependent variable and, for the models of

this study, will also control endogeneity for the credit portfolio sales variables.

After estimating the models, instruments overidentification and autocorrelation in

first-difference errors were tested to make sure the models utilized obey the

restrictions of non autocorrelation and non overidentification. For instrumental

variables overidentification, it was used the Sargan Test and for autocorrelation,

Arellano-Bond test for zero autocorrelation in first-differenced erros. All the tests

showed the models were adequate, allowing the results usage.

4.5..1 Dependent Variables

Rating and Capital Adequacy

The proxy for the portfolio rating was built based on a credit default provision –

PCLD - stipulated by Brazilian law: Resolution 2682. This Resolution describes how

a default provision has to be calculated considering credit operations ratings. The

provisions are expected values based on the following default probability table:

Table II: Credit Rating

Default probabilities according to loans credit rating defined at the Banco Central do Brasil Resolution 2682

Credit

Rating

Default

Probability

AA 0%

A 0,5%

B 1%

C 3%

D 10%

E 30%

F 50%

G 70%

H 100%

And the calculus formula is:

( )

∑

AAi

ValueOperationCredit*yProbabilitDefaultPCLD

(1)

Where:

i is the portfolio rating – H to AA;

=PCLD is the Brazilian term for credit default provision;

yProbabilitDefault is the probability – according to the table above - applied to

operations classified as i;

ValueOperationCredit is the volume of credit operations classified as i

Instead of a default provision, we are interested in quality classification (rating).

As the expected signs between a default provision and a portfolio rating are

opposites, the proxy utilized here will be:

ValueOperation Credit Total

PCLD

Rate

−

(2)

It is important to highlight that, once “Total Credit Operation Value” involves the

entire credit rating portfolio, differently from PCLD, this proxy considers AA portfolios

for rating estimation.

As we also want to test if capital adequacy does or does not make its rating

worse, the capital adequacy ratio – CAR - will be one of the independent variables

regressed against Rating.

Relative Loans Volume Change

As anticipated on session 4.1, according to the resolution 2682, there are nine

groups of credit rating, classified by letters from AA to H, where AA is the best payer

group and H is the worst one. The resolution also determines the highest

classification a portfolio may have, given a certain delay from the borrowers on

paying its debts:

Rule Rating

delay between 15 and 30 days risk level B

delay between 31 and 60 days risk level C

delay between 61 and 90 days risk level D

delay between 91 and 120 days risk level E

delay between 121 and 150 days risk level F

delay between 151 and 180 days risk level G

delay over 180 days risk level H

There are other issues, such as economic situation and leveraging, that are

considered on the classification and may promote its downgrade once the levels are

the highest possible for such portfolios.

For this study, the maximum classification will help on ratings groups division. As

portfolios with credit rating AA and A have the same “no delay” characteristic, they

will be observed together, as the best payers group. Following the idea of merging

similar groups, B and C became a group with delay between 15 and 60 days; D and

E between 61 and 120 days; and F, G and H over 121 days. As financial institutions

were classified in four groups (quarters), these groups from now on will be called,

from the best to the worst payers respectively, FirstQ, SencondQ, ThirdQ and

FourthQ.

To check credit portfolio sales influence over portfolio growth, the Relative Loans

Change on those categories volumes were calculated by the following formula.

Loans

ChangeLoans Total

ChangeLoans

RLC

Q1j

1tj,

Q1j

tj,

1tj,

tj,

−













=−=

∑

−

(3)

Where:

j is the merged portfolio group – FirstQ to FourthQ;

tj,

RLC

is the relative loans change on group j at time t;

Profitability

Profitability will be represented by Return on Equity:

Equity

IncomeNet

ROEityProfitabil ==

(4)

4.5..2 Independent Variables

Credit portfolio sales

The credit portfolio sales - variables of interest - were represented by credit

portfolio, being classified into eight sales groups:

• rlfi: Credit portfolio sales with recourse done to linked financial institutions;

• rnlfi: Credit portfolio sales with recourse done to non-linked financial

institutions;

• rls: Credit portfolio sales with recourse done to linked securitizers;

• rnls: Credit portfolio sales with recourse done to non-linked securitizers;

• lfi: Credit portfolio sales without recourse done to linked financial

institutions;

• nlfi: Credit portfolio sales without recourse done to non-linked financial

institutions;

• ls: Credit portfolio sales without recourse done to linked securitizers;

• nls: Credit portfolio sales without recourse done to non-linked securitizers.

According to the Institutions Accounting Plan of Brazilian Finance System –

COSIF - linked institutions or securitizers are companies affiliated, controlled or

controllers as well as companies that, through direct or indirect common control, are

included in the same financial conglomerating or the economic-financial institution.

Capital Adequacy Ratio (CAR)

CAR is the Basel Capital Adequacy Ratio, in Brazil it is also called Basel Index,

and its calculus is done by the following formula:

Assets tedRisk Weigh

Capital IITier Capital ITier

= CAR

(5)

Where:

Tier I Capital is the most permanent and readily available support against

unexpected losses;

Tier II Capital is not permanent in nature or, is not readily available;

RWA, Risk Weighted Assets represents the sum of the assets weighted by their

respective risks.

Whether CAR involves this calculus – and others even more complex behind TIER I

and TIER II definitions, they will not be demonstrated, because for this study it was

not calculated but obtained done.

4.5..3 Control Variables

For all models it will be necessary to use some control variables to make the

results reflect only the analyzed variables effects, taking out of the model external

influences. The control variables, are the ones usually adopted by the literature and

will be: size, interest rating, portfolio quarterly growth, assets quarterly growth,

liquidity, profitability quarterly growth and capitalization.

These variables influence the decision to sell loans portfolios, as well as their

amounts. Financial institutions would check them before deciding what kinds of credit

portfolio sales will be done and how much money the operation will involve.

Therefore, the control variables are of one quarter before the quarter analyzed and to

do that in the models, lags of them will be used. For each model, according to the

dependent variable, different combinations of variables will be used in a way that

makes good interpretative sense. In summary, the proxies for all the control variables

utilized were:

• l.Size

Represents the lagged total assets on the balance sheet at the end of the

quarter as a proxy for the financial institution’s size.

• l.Interest Rating

Represents the lagged CDI, the Brazilian interbanks deposits interest rating at

the end of the quarter

• l.Portfolio Quartely Growth

Represents the lagged changes between quarters on the financial institutions’

credit portfolio. Defined as the difference between the value of the total credit

portfolio of the preview quarter and the total credit portfolio of the quarter

before the preview one, divided by the total credit portfolio of the quarter

before the preview one.

• l.Assets Quartely Growth

Represents the lagged changes between quarters on the financial institutions’

assets. Defined as the difference between the value of the total assets of the

preview quarter and the total assets of the quarter before the preview one,

divided by the total assets of the quarter before the preview one.

• l.Liquidity

Represents a proxy for financial institution’s liquidity. Defined as the lagged

sum of: 1)Cash and Cash Equivalents, 2) Securities and Derivative Financial

Instruments and 3) Interbank Accounts, divided by total assets on the balance

sheet at the end of the quarter.

• l.Quarterly Profitability Growth

Represents the lagged changes between quarters on the financial institutions’

return on equity - ROE. Defined as the difference between the value of the

ROE of the preview quarter and the ROE of the quarter before the preview

one, divided by the ROE of the quarter before the preview one.

• l.Capitalization

Represents the lagged total equity divided by total assets on the balance

sheet at the end of the quarter as a proxy for the financial institution’s

capitalization.

4.5..4 Descriptive Statistics

Table III – Statistics of Sales of Credit Portfolio

This table reports the credit portfolio sales frequency distribution by Linkage and Institution separated by the existence or not of

recourse on the operations for the whole sample of credit portfolio sales originated by 145 Brazilian financial institutions from

the first quarter of 2001 to the second quarter of 2008. The third and fifth, fourth and the sixth rows present the values and

percentages over the registered quarters when SPVs negotiated credit portfolio and total volumes of credit portfolio sales,

respectively. Except for the number of registered quarterly negotiated, all of the other descriptive statistics are represented in

millions of reais.

Negotiated

Quarters

Volumes of

Sales

Credit Portfolio Sales Statistics

Intitution

(Buyer) Recourse

Num

% Value %

Mean

Million

Median

Million R$

Min

Million R$

Max

Million

St. Dev

Million R$

With

1.03%

1805

1.59%

150

0.003

590

248

Linked

Financial

Istitutions

Without

2.62%

6094

5.37%

265

119

7.778

1188

337

With

517

35.66%

40326

35.52%

1344

509

179.508

4182

1319

Non-Linked

Financial

Intitutions

Without

272

18.76%

33414

29.43%

1152

929

6.526

2798

976

With

4.21%

542

0.48%

4.099

Linked

Securitizers

Without

258

17.79%

25513

22.47%

850

780

3.860

3337

726

With

103

7.10%

535

0.47%

0.219

Non-Linked

Securitizers

Without

186

12.83%

5293

4.66%

176

3.065

874

216

Total

1450

113521

In table III, it is possible to observe that the number of negotiated quarters and

the total volume of receivable sales done to non-linked financial institutions are much

higher than sales to other institutions and that, once they are done to non-linked

financial institutions, they are mostly done with recourse. In addition, for credit

portfolio sales done to other institutions, the behavior found was different; they are

mostly done without recourse, which may mean that non-linked financial institutions

are stricter than other institutions when buying loans, asking, for the most part of their

negotiations, financial institutions to assume part of the risk sold. The mean, median,

minimum and maximum values of receivable sales done to non-linked financial

institutions are also the highest ones and, as there are more data for this type of

sales, its standard deviations are also shown to be the highest ones.

Linked securitizers seem to be the second mostly involved institution and have

relatively high number of negotiated quarters, sales volumes and, especially for

operations without recourse, relatively high mean, median, minimum, maximum and

standard deviation. On other hand, analyzing number of negotiated quarters and

volume of sales, linked financial institutions do not appear to be a big receivable

buyer.

Compared to their total purchases, securitizers buy relatively low amounts of

receivables with recourse having their biggest part of operations done without it,

which indicates that, differently from non-linked financial institutions, those institutions

are not too strict when buying credit portfolios.

5 RESULTS

5.1 Rating and Capital Adequacy

Table IV – Sales of Credit Portfolio, Rating and Capital Adequacy

Dependent Variable: Rating of the credit portfolio

The ratings of the credit portfolios are analyzed using Arellano and Bond panel data analysis, controlling receivable sales

endogeneity. The portfolio rating, calculated by dividing the Resolution 2682 Default Provision by the negative value of the total

credit operation, is the dependent variable. Independent variables - Credit portfolio sales: rlfi: with recourse to linked financial

institutions; rnlfi: with recourse to non-linked financial institutions; rls: with recourse to linked securitizers; rnls: with recourse to

non-linked securitizers; lfi: without recourse to linked financial institutions; nlfi: without recourse to non-linked financial

institutions; ls: without recourse to linked securitizers; nls: without recourse to non-linked securitizers. CAR is the Basel measure

of capital adequacy. Interest rating, portfolio growth, liquidity and ROE growth are independent control variables.

Coef.

Independent Variables

l.1

df.rating 0.3244

*** 9.68

rlfi -2.58E-05

-0.52

rnlfi -9.02E-09

-0.85

rls -5.34E-08

-0.09

rnls 2.93E-06

*** 2.64

lfi 6.78E-08

* 1.93

nlfi 4.57E-09

1.44

ls -9.37E-09

-1.24

nls 4.53E-08

0.35

l.size -3.72E-14

-1.17

l. interest_rating 0.046154

*** 15.77

l.portfolio_ growth 0.004215

*** 3.19

l.liquidity 0.002548

0.31

l.ROE_growth 4.67E-10

*** 6.52

l.car -0.000323

*** -12

cons -0.561757

*** -17.38

*** Significant at the 1% level; ** Significant at the 5% level; * Significant at the 10% level

As shown in table IV, only two of the eight credit portfolio sales groups have a

direct influence over the remaining portfolio rating: rnls – credit portfolio sales with

recourse done to non-linked securitizers and lfi – credit portfolio sales without

recourse done to linked financial institutions. They had positive relations pointing that

financial institutions do these types of sales looking for risk mitigation. The portfolio

rating improvement, promoted by rnls and lfi, may be seen as an indicator that

financial institutions usually prefer selling, to those groups, bad rated portfolios and

and that these operations provide an upgrade on the remaining portfolio.

CAR showed to be significant and its relation to Rating was negative, which

means that the more capital adequate a financial institution is, the worse its credit

ratio will be. This result means that financial institutions are taking more risky loans

once they reach better capital adequacy to compensate the expected profit losses

traded.

Although the rating model had significant relations for only two types of credit

portfolio sales, the loans volume change and profitability models presented many

coherent results that will be described in the following sections.

5.2 Loans Volume Change

Along with other control variables, volumes of all categories of credit portfolio

sales were regressed against four categories of relative loans changes to check the

influence of recourse, linkage and type of institution on each of the loan portfolio

categories: FistQ, SecondQ, ThirdQ and FourthQ, ranked according to credit quality.

There is a way of interpreting the relation of credit portfolio sales groups with

loans changes given a specific portfolio: A negative relation means that the

institutions that have the credit portfolio sales group receive loans of the specific

portfolio, with or without recourse. For example: SecondQ and rls had a negative

relation. This means that linked securitizers are buying SecondQ with recourse. As

loans changes are relative to the whole portfolio change, the positive sign only

means that when credit portfolio sales are done to that sale group, the operations are

not of that specific portfolio type but of any other type.

Negative evidence is stronger when concluding about risk mitigation or growth

than positive evidence. This is justified because, in the first case, as there are first

differences working as instruments to control loans changes endogeneity and

portfolio growth working as a control variable for the whole portfolio change influence,

it is possible to assert that, in fact, there was a credit sale of the same quality as the

loans change type. The endogeneity instrument and the portfolio control variable

guarantee that the interpretation of credit portfolio sales groups relation are only

about credit portfolio sales.

On the other hand, positive signs are weak evidence because, given this

controlled scenario, a growth on loans changes indicates only that some other kind of

loan is being sold, not informing more about the portfolio involved in the operation.

Table V – Sales of Credit Portfolio and Remaining Portfolios Structure

Dependent Variable: Change on relative loans volume

The relative changes on loans volumes – dependent variables - were analyzed using Arellano and Bond panel data

analysis, controlling for receivable sales endogeneity. The dependent variables were calculated by finding each group quarterly

loans volume change, relative to the total portfolio change. The loans change groups were: FisrtQ, that assembles AA and A

portfolios; SecondQ, B and C portfolios, ThirdQ, D and E portfolios and FourthQ , F, G and H portfolios. Independent variables -

Credit portfolio sales: rlfi: with recourse to linked financial institutions; rnlfi: with recourse to non-linked financial institutions; rls:

with recourse to linked securitizers; rnls: with recourse to non-linked securitizers; lfi: without recourse to linked financial

institutions; nlfi: without recourse to non-linked financial institutions; ls: without recourse to linked securitizers; nls: without

recourse to non-linked securitizers. Interest rating, portfolio growth, liquidity, ROE growth and capitalization are independent

control variables.

Portfolio Relative Change

FirstQ SecondQ ThirdQ FourthQ

Independent Variables

Coef.

l.1

df.port_relative_chg -10.6124

*** -0.31886

*** -8.33E-07

-2.20754

***

rlfi 0.00582

6.46E-05

0.000464

0.114801

rnlfi -0.00024

*** -9.33E-07

*** 1.30E-07

0.003186

***

rls -0.006515

*** -0.00012

*** 0.000108

*** 0.018132

***

rnls 0.000693

-2.3E-05

*** 0.000151

*** -0.01163

***

lfi 0.000053

-6.94E-07

*** -2.43E-06

-0.00171

***

nlfi 4.21E-05

** 6.76E-07

*** 8.17E-06

*** 0.001044

***

ls 4.13E-05

*** 2.83E-07

*** -8.23E-07

*** -0.0005

***

nls 0.000182

4.94E-06

*** 6.31E-06

* -0.00099

***

l.size 1.74E-10

6.22E-12

*** -9.44E-12

*** 2.05E-09

l. interest_rating -3.592617

-0.97942

*** 0.443615

*** 498.4105

***

l.portfolio_ growth -0.154712

0.011909

0.088466

* -43.1787

***

l.liquidity 153.2244

** -2.4594

*** -14.2617

*** -717.865

***

l.ROE_growth 7.12E-08

-1.59E-08

2.66E-08

1.72E-06

l.capitalization 0.788571

0.106145

0.007617

-15.4191

cons -19.22616

5.783076

*** 3.982933

*** -1925.89

***

*** Significant at the 1% level; ** Significant at the 5% level; * Significant at the 10% level

According to table V, at the FirstQ model, rnlfi and rls are shown to have a

negative relation to the best rated portfolio while nlfi and ls showed a positive

relation. For the negative relations, there is strong evidence that financial institutions

are selling the best credit rating loans portfolios to non-linked financial institutions

and to linked securitizers, both with recourse. For positive relations, there is weak

evidence that when non-linked financial institutions and linked securitizers buy loans

portfolios without recourse, they do not buy the best type of portfolio.

These results present an interesting behavior: even though the selling of good

portfolios means risk increase, all the credit portfolio sales groups that were shown to

be negatively significant were “with recourse” groups. This indicates that for the

cases where leveraging is done, they are done with recourse and financial

institutions are worried about something besides leveraging: the remaining portfolio

rating; keeping part of this good portfolio ceded at their balance sheets.

At the SecondQ model, rnlfi, rls, rnls and lfi showed to be negative related to the

second best rated loans group while nlfi, ls and nls were shown to be positively

related. For the negative relations, there is strong evidence that financial institutions

are selling the second best credit rating loans to non-linked financial institutions,

linked and non-linked securitizers, all of them with recourse and to linked financial

institution without recourse. For positive relations, there is a weak evidence that

when non-linked financial institutions, linked securitizers and non-linked securitizers

buy loans portfolios, all of them without recourse, they do not do it for the second

best rated portfolio group.

Most part of these results for SecondQ, assuming that portfolios that compose

SecondQ (B and C) are good enough to promote good earnings by sales, reinforce

the behavior of leveraging with recourse found on the model for FirstQ.

At the ThirdQ model, lfi and ls are shown to be negatively related to it while rls,

rnls, nlfi and nls showed to be positively related. For the negative relations, there is

strong evidence that financial institutions are selling the ThirdQ portfolio loans –

portfolios D and E - to linked financial institutions and securitizers, both without

recourse. For positive relations, there is weak evidence that when credit portfolio

sales to linked and non linked securitizers, both with recourse, and non-linked

financial institutions and securitizers, both without recourse, happens, they do not do

it using the ThirdQ group.

Considering that ThirdQ is a poorly rated group, differently from FirstQ and

SecondQ, its sale represents risk mitigation. Therefore, these results would point with

strong evidence that ls and lfi – sales to linked companies without recourse - are

done for risk mitigation.

At the FourthQ model, rnls, lfi, ls and nls are shown to be negatively related to it

while rnlfi, rls and, nlfi are shown to be positively related. For the negative relations,

there is strong evidence that financial institutions are selling the worst loans –

portfolios F, G and H - to linked financial institutions and securitizers, both without

recourse, and to non-linked securitizers with and without recourse. For positive

relations, there is weak evidence that when credit portfolio sales to linked securitizers

with recourse and non linked financial institutions, with or without recourse, happen,

the worst type of portfolio is not used to do it.

Given that the FourthQ is the worst credit rating portfolio, its sale surely

represents risk mitigation. Therefore, these results point with strong evidence that

rnls, lfi, ls and nls are done for risk mitigation. Between these risk mitigation groups,

only one is with recourse. Except for rls – linked with recourse - in general,

securitizers promote risk mitigation by buying FourthQ.

By putting all four models of information together in table VI, we may have a

better view of the credit portfolio sales institutions’ behavior.

Analyzing linked securitizers, we see that, when they buy good loans portfolios to

foment leveraging, they do it with recourse. For the cases without recourse, the

evidence suggests risk mitigation when the worst rated loans are bought on this

condition.

The non-linked securitizers buy the worst rated loans, independent of if it is done

with or without recourse and, despite buying SecondQ with recourse, which would

indicate leveraging, they do not buy firstQ and are characterized as “risk mitigaters”.

Looking at the linked financial institutions, it is important to highlight that their

operations with recourse were not significant for any of the groups, which may

indicate that they are not done with a specific objective. When loans are bought by

these institutions without recourse, only one type is not operated: FirstQ, showing

that they are just not buying the best portfolio, promoting risk mitigation. Those

results reinforce the results found at the rating model, which pointed to linked

financial institutions as “risk mitigaters”.

The last group of institutions is the non-linked financial institutions. Their results

show that they help financial institutions with leveraging but, just like linked

securitizers, they always do so by buying loans portfolios with recourse. It is not clear

whether the cases without recourse promote leveraging or risk mitigation, because

there were no negative relations between the loans sale group and any of the

portfolio change variables. Their relations for all the models are positive, which may

indicate that this type of credit portfolio sales do not belong to a single group but

probably to a combination of them. The following table is a summary of the relations

for the four models:

Table VI: Summary of Empirical Evidence on Change in Portfolio

Hypotheses of Mitigation and Leveraging were empirically tested through Arellano and Bond panel data models, controlling

receivable sales change. Brazilian Financial Institutions’ quarterly data between March 2001 and September 2008 were used.

Found signs for the credit portfolio sales groups are on columns two to five according to loans changes categories. Brief

Comments are on the sixth column, being restricted to negative signs – the ones that represent strong Evidence. Hypotheses

supported by the results are on the seventh column.

Credit

portfolio

sales

Groups

Portf.

Signs

Portf.

Signs

Q Portf.

Signs

Portf.

Signs

Brief Comments

(Strong Evidence)

Supporting

Hypotheses

Rlfi

No evidence was found …..

None

Rnlfi

- - + Buy the two best portfolios Leveraging

Rls

- - + + Buy the two best portfolios Leveraging

Rnls

- + - Buy the worst and the second

best portfolios, but do not buy

the best one

Risk Mitigation

Lfi

- - - Just do not buy the best portfolio Risk Mitigation

Nlfi

+ + + + Indefinite results ….. …

None

+ + - - Buy the two worst portfolios Risk Mitigation

Nls

+ + - Buy only the worst portfolio Risk Mitigation

The same model was applied to the total portfolio change to check if, at the end,

the resources earned by credit portfolio sales are making the total portfolio bigger or

smaller. If they have a positive relation and are making the total portfolio bigger, they

are using credit portfolio sales to for leveraging. On the other hand, if they have no

significance or negative relation, the behavior is of maintenance or reduction on the

total portfolio size, which indicates risk mitigation. At this model and the others the

portfolio growth variable was replaced by an assets growth variable.

Table VII shows overall changes on loans volume. As there is an instrument for

the dependent variable first difference lags, the portfolio growth, a high correlated

control variable, was replaced by the assets growth.

Table VII – Sales of Credit Portfolio and Total Remaining Portfolio

Dependent Variable: Change on total loans volume

The changes on total loans volume – dependent variable - were analyzed using Arellano and Bond panel data analysis. The

dependent variable was calculated as following:

−

−tt

LoansLoans

; where

Loans

represents the whole portfolio

value at time t. Independent variables - Credit portfolio sales: rlfi: with recourse to linked financial institutions; rnlfi: with recourse

to non-linked financial institutions; rls: with recourse to linked securitizers; rnls: with recourse to non-linked securitizers; lfi:

without recourse to linked financial institutions; nlfi: without recourse to non-linked financial institutions; ls: without recourse to

linked securitizers; nls: without recourse to non-linked securitizers. Interest rating, assets growth, liquidity and ROE growth are

independent control variables.

Coef.

Independent

Variables

l.1

df.total_port_chg -0.06692

***

-5.77

rlfi 5.05E-05

0.4

rnlfi -8.59E-08

-1.04

rls 3.16E-06

0.89

rnls -8.48E-07

** -2

lfi 1.38E-08

0.07

nlfi -4.36E-07

***

-14.11

ls -7.65E-08

***

-6.92

nls 8.06E-08

0.64

l.size -3.69E-12

***

-13.39

l. interest_rating -0.00805

-0.77

l.assets_ growth 0.010833

***

7.4

l.liquidity -0.053545

***

-2.91

l.ROE_growth 3.34E-09

** 2.1

l.capitalization 0.000116

0.14

cons 0.274708

***

4.85

*** Significant at the 1% level; ** Significant at the 5% level

According to table VII, all sales groups that are shown to be significant for loans

change – rnls, nlfi and ls - had negative relations to it. This not only means that credit

sales, done to these groups, do not promote new waves of loans but also indicates a

relative reduction of operations when compared to the preview quarter, supporting

the risk management hypothesis.

Non Linked Finance Institution with recourse –nlfi -, a group that so far have not

had strong evidence for any of the hypotheses, for this last model, showed to be a

“risk mitigater”

5.3 Profitability

To reinforce the results found with the prior rating and loans changes models, it

was tested whether credit portfolio sales indicate an improvement or reduction in

profitability. As presented in table I for credit portfolio sales done by financial

institutions with leveraging aim, there is an expected improvement in profits and,

therefore, a positive relation between groups that promote leveraging and

profitability.

Another assertion may be made about negative relations and risk mitigation:

when financial institutions sell loans and do not use the funds earned via risky

activities, they reduce profits from interest and therefore become less profitable. The

following table displays these reinforcing results.

For this model, as there are instruments for the dependent variable first difference

lags, ROE growth, a high correlated control variable was not used.

Table VIII shows that all credit portfolio sales that are shown to be used for

leveraging in the prior models – to linked securitizers and non-linked financial

institutions, both with recourse - had positive relations to profitability; reinforcing that,

in these cases, leverage is being done to improve profitability instead of to manage

risks.

Linked financial institutions, as detected in the prior models, are used by financial

institutions to promote risk mitigation beyond without recourse operations. On table

VIII this behavior was also found, once there was a negative relation between lfi and

profitability, proving that the risk mitigation done in these circumstances is also

reducing profits.

Non-linked financial institutions without recourse, the group that was only shown

to have a risk mitigation proposal, also had a negative relation to profitability and was

shown once more to be a “profit reductor“, which reinforces the risk management

behavior found.

Between credit portfolio sales to linked securitizers, only the ones with recourse

were significant and, as already described, positively related to profitability.

Credit portfolio sales with recourse to non-linked securitizers, as was pointed out

by the previous models, indicate a portfolio’s risk management, presenting a negative

relation with profitability. For operations to the same institutions, but without recourse,

there was a positive relation, indicating that even with a risk mitigation proposition

this type of credit portfolio sales promotes profitability. This last relation was the only

opposite behavior from what was expected and did not reinforce the prior results.

Table VIII – Sales of Credit Portfolio and Profitability

Dependent Variable: ROE

The dependent variable Return on Equity - ROE – was analyzed using Arellano and Bond panel data analysis. and was

calculated as following:

Equity/IncomeNet

. Independent variables - Credit portfolio sales: rlfi: with recourse to linked

financial institutions; rnlfi: with recourse to non-linked financial institutions; rls: with recourse to linked securitizers; rnls: with

recourse to non-linked securitizers; lfi: without recourse to linked financial institutions; nlfi: without recourse to non-linked

financial institutions; ls: without recourse to linked securitizers; nls: without recourse to non-linked securitizers. Interest rating,

portfolio growth, liquidity and capitalization are independent control variables.

Coef.

Independent Variables

l.1

df.profitab -0.69807

*** .

rlfi -36.3486

-0.21

rnlfi 2.485664

*** 20.27

rls 336.5492

*** 17.96

rnls -6.06743

** -2.42

lfi -0.96597

*** -3.55

nlfi -1.05931

*** -11.5

ls 0.015862

0.46

nls 1.172699

** 2.53

l.size 1.93E-06

*** 9.31

l. interest_rating -3161286

*** 3417.52

l.portfolio_ growth -567.441

*** -5.16

l.liquidity 2672043

*** 64.31

l.capitalization 84.53031

0.96

cons 15300000

*** 145.78

*** Significant at the 1% level; ** Significant at the 5% level; * Significant at the 10% level

6 CONCLUSIONS

Contrarily of the US and Canadian banks behavior found by Cebenoyan and

Strahan (2001) and Dionne and Harchaoui (2003) respectively, Brazilian financial

institutions usually do not use credit portfolio sales as a leveraging instrument. When

they do it, leveraging themselves by selling the best portfolios, they do it with

recourse, leaving part of the portfolio sold on their balance sheets. This helps to keep

a good credit rating for the remaining portfolio and, consequently, do not promote

expressive changes on assets risks. These results drive us to the conclusion that

sales of credit portfolios mitigate risks, supporting the results found by Staton(1998)

and Murray(2001)

Despite of being worried about keeping a good remaining credit portfolio rating,

financial institutions actually do a relatively high number of good portfolios sales. The

non-linked ones trade this type of operation in higher volumes than linked securitizers

the other type of institutions that also does it.

The other companies - non-linked securitizers and linked financial institutions,

both subdivided in with and without recourse – only promote risk mitigation, what

reinforces the idea that, similarly to Thomas and Wang(2004), according to the

securitization vehicle and the existence of recourse, there are different purposes on

dealing credit portfolio sales.

Between the groups that promoted risk mitigation, only linked securitizers buy low

rated portfolios with recourse. The other ones only do it without recourse, helping the

financial institutions to completely remove high risk assets from their balance sheets.

The more adequate Brazilian financial institutions are, those having the highest

ratios for CAR, the Basel Capital Adequacy Ratio, the more likely they will be to

have lower credit rating portfolios, compensating profit losses, implicit in the

adequacy - by profit gains - implicit in the higher interest of bad rated loans. This

result supports Dionne and Harchaoui(2003) founds about capital adequacy for

Canadian banks.

7 REFERENCES

ARELLANO, M.; BOND; S. Some Specification Tests for Panel Data: Monte Carlo

Evidence and an Application to Employment Equations. Review of Economic Studies

58, 277–298, 1991.

Banco Central do Brasil. Resolução 2682. Electronic copy available at

www.bcb.gov.br

BREWER III, Elijah; MINTON, Bernadette A.; MOSER, James T. Interest-rate

Derivatives and Bank Lending, Journal of Banking and Finance 24, 353-379, 2000.

BANNIER, Christiana E.; HÄNSEL, Dennis N. Determinants of Banks´ Engagement

in Securitization: Banking and Financial Studies, Discussion Paper Series 2. October,

2008. Electronic copy available at:http://ssrn.com. Accessed on April, 2009.

CATÃO, Gustavo; RODRIGUES, Raimundo N.; LIBONATI, Jeronymo J.

Securitização de Recebíveis no Setor Bancário Brasileiro: Um Estudo Empírico. VIII

Congresso USP de Controladoria e Contabilidade. July, 2008.

CEBENOYAN, Sinan A; STRAHAN, Philip E. Risk Management, Capital Structure

and Lending at Banks. Working Paper, The Wharton School, University of

Pennsylvania, 2001.

DIONNE, Georges; HARCHAOUI, Tarek M. Banks’ Capital, Securitization and Credit

Risk: An Empirical Evidence for Canada. Les Cahiers du CREF. May, 2003.

Electronic copy available at: http://ssrn.com. Accessed on april, 2009.

Statement of Cameron L. Cowan Partner Orrick, Herrington, and Sutcliffe, LLP.

Hearing on Protecting Homeowners: Preventing Abusive Lending While Preserving

Access to Credit. American Securitization Forum. November, 2003.

FABOZZI Jr., Frank J.; DAVIS, Henry A.; CHOUDHRY, Moorad. Introduction to

structured finance. New Jersey: John Wiley Trade, 2006.

HENDERSON, John; SCOTT, Jonathan P. Securitization, New York Institute of

Finance, New York 1998.

KASHYAP, Anil K., RAJAN, Raghuram G.; STEIN, Jeremy C. Banks as Liquidity

Providers: An Explanation for the Coexistence of Lending and Deposit-Taking.

Journal of Finance, Vol. 57, pp. 33-73. 2002 .

MINTON, Bernadette; SANDERS, Anthony B. STRAHAN, Philip E. Securitization by

Banks and Finance Companies: Efficient Financial Contracting or Regulatory

Arbitrage? Working Paper. October, 2004. Accessed on April, 2009.

MURRAY, Allan P. Has Securitization Increased Risk to the financial System?

Bussines Economics, January 2001.

PAVEL, CRISTIANE A. Securitization. The Analysis and Development of the Loan-

Based/Asset Backed Securities Markets. Probus Publishing. Chicago, Illinois, 1989.

PINHEIRO, FERNANDO A. P. Securitização de Recebíveis – Análise dos Riscos

Inerentes. Dissertação (Mestrado em Administração). Programa de Pós-Graduação

em Administração - Faculdade de Economia, Administração e Contabilidade da

Universidade de São Paulo, 2008.

PLANO CONTÁBIL DAS INSTITUIÇÕES DO SISTEMA FINANCEIRO NACIONAL –

COSIF. Capitulo 1: Normas Básicas 1, Sessão 1: Princípios Gerais Item 9:

Sociedades Ligadas.

SECURATO, JOSÉ R., Cálculo Financeiro das Tesourarias :Bancos e Empresas,

São Paulo, 2002.

STANTON, Sonya W. The Underinvestment Problem and Patterns in Bank Lending.

Journal of Financial Intermediation Vol. 7, No. 3. 1998.

THOMAS, Hugh; WANG, Zhiqiang. Banks Securitization and Risk Management.

Draft submitted for review to the Journal of Money, Credit and Banking. June, 2004.

Electronic copy Available at: http://ihome.cuhk.edu.hk. Accessed on April, 2009

TOMIATTI, Claudio R., DE OLIVEIRA, Edson R. Mercado de capitais: Securitização.

Revista da Pós-Graduação, Vol. 1, No 2 (2007), Unifeo

WOOLDRIGE, Jeffery M. Econometric Analysis of Cross Sections and Panel Data.

Cambridge, Mass. : Massachusetts Institute of Technology, 2001.

FUNDAÇÃO GETULIO VARGAS

ESCOLA DE ADMINISTRAÇÃO DE EMPRESAS DE SÃO PAULO

GRAZIELLA LAGE LAUREANO

SALE OF CREDIT PORTFOLIO AND RISK: THE CASE OF FINANCIAL

INSTITUTIONS IN BRAZIL

SÃO PAULO

2009

GRAZIELLA LAGE LAUREANO

SALE OF CREDIT PORTFOLIO AND RISK: THE CASE OF FINANCIAL

INSTITUTIONS IN BRAZIL

Dissertação apresentada à Escola de

Administração de Empresas de São Paulo da

Fundação Getúlio Vargas, como requisito para

a obtenção de título de Mestre em

Administração de Empresas

Campo de Conhecimento:

Mercados Financeiros e Finanças Corporativas

Orientador:

Prof. Dr. Richard Saito

SÃO PAULO

2009

Laureano, Graziella.

SALE OF CREDIT PORTFOLIO AND RISK: THE CASE OF FINANCIAL

INSTITUTIONS IN BRAZIL / Graziella Lage Laureano. - 2009

42 f.

Orientador: Richard Saito.

Dissertação (mestrado) - Escola de Administração de Empresas de São

Paulo.

1. Título de créditos -- Brasil. 2. Instituições financeiras -- Administração.

3. Securitização. 4. Risco (Economia). I. Saito, Richard. II. Dissertação

(mestrado) - Escola de Administração de Empresas de São Paulo. III. Título.

CDU 336.763(81)

GRAZIELLA LAGE LAUREANO

SALE OF CREDIT PORTFOLIO AND RISK: THE CASE OF FINANCIAL

INSTITUTIONS IN BRAZIL

Dissertação apresentada à Escola de

Administração de Empresas de São Paulo da

Fundação Getúlio Vargas, como requisito para

obtenção de título de Mestre em Administração

de Empresas

Campo de Conhecimento:

Mercados Financeiros e Finanças Corporativas

Orientador: Prof. Dr. Richard Saito

Data da Aprovação:

_____ / _____/ ________

Banca Examinadora:

Prof. Dr. Richard Saito (Orientador)

FGV – EAESP

Prof. Dr. Rodrigo De Losso

FGV – EAESP

Prof. Dr. Roberto Calfat

IBMEC - SP