Disentangling the impact of securitization on bank profitability

TL;DR: In this article, the role played by bank risk, cost of funding, liquidity and regulatory capital in explaining the relationship between securitization and bank profitability was analyzed, and the contribution of each individual transmission channel in the overall impact on bank profitability were identified.

About: This article is published in Research in International Business and Finance.The article was published on 2019-01-01 and is currently open access. It has received 15 citations till now. The article focuses on the topics: Capital requirement & Market liquidity.

Summary

1 Introduction

• Securitization has fundamentally altered the way in which financial intermediation is organized as it has provided banks with various incentives to improve efficiency and performance.
• In the aftermath of the 2008 credit crisis, securitization markets became subject to intensive regulatory reforms which implied the curbing of certain higher risk activities.
• The authors add to these studies by integrating both channels of securitization effects and further investigating their simultaneous impact on bank profitability.
• In addition, the authors show that bank risk and cost of funding have positive impact on bank profitability.
• Sections 3 provides an overview of related literature and develops the research hypotheses.

2 Developments in the US Banking Industry and Securitization Market

• The US banking industry has experienced an enormous transformation over the course of the last few decades.
• This trend has fundamentally altered the risk-return profiles of US banks over the last few decades (DeYoung and Roland, 2001) .
• Particularly, banks costs of production were static or declining and there has been an increase in total revenues from traditional and non-traditional sources.
• Indeed, until mid-2007 it was widely perceived that the US banking system was sound and performing well, particularly because banks capital holdings and profitability appeared to be high and at record levels.
• Securitization activities played a pivotal role for the housing market in the run up to the credit crisis of 2008 as the Asset Backed Securities (ABS) and covered bonds provided between 20 and 60 per cent of the funding for new residential mortgage loans originated in mature economies (IMF, 2009) .

3 Literature Review and Hypotheses Development

• This paper is closely related to the strand of literature that studies the impact of securitization on bank profitability.
• Moreover, proceeds from securitizations can be reinvested in loans directed to new profitable projects, thus aligning the average rate of the bank's loan portfolio with the market rate and increasing the bank's income from interest (Thomas, 1999) .
• They show that profitability is significantly and positively affected by securitization.
• The mechanism of securitization activities implies that it has implications for bank performance.
• In the subsections below, the authors develop the necessary hypotheses to test this argument.

3.1 Bank Risk Channel

• On the one hand, it may reduce bank risk by shifting credit risk to the market and improving risk sharing opportunities.
• Ambrose et al. (2005) suggest a positive effect of securitization on bank credit risk due to retaining riskier loans while selling safer ones in response to regulatory requirements.
• The predominant type of bank risk is credit risk which materializes when a loan becomes irrecoverable, or a borrower fails to meet the loan servicing costs in time.
• Moreover, Athanasoglou et al. (2008) report that excess exposure to credit risk reduces profitability.
• Nevertheless, Tan (2016) show that the impact of risk on bank profitability is insignificant especially when considering the impact of banking industry competition.

3.3 Liquidity Channel

• The traditional incentives of securitization entail increasing liquidity as a primary objective because securitization allows banks to liquidate illiquid assets (Cardone-Riportella et al., 2010) .
• This view ignores the role of securitization in managing liquidity risk, while emphasizing the increase in bank liquidity based on traditional measures of liquidity.
• Moreover, Affinito and Tagliaferri (2010) show that securitization is used by banks to improve liquidity positions and to mitigate liquidity risk exposures.
• Banks may decide to hold liquid assets to reduce risks and to avoid bank failures (Imbierowicz and Rauch, 2014) .
• While Bourke (1989) and Pasiouras and Kosmidou (2007) shows a significantly positive relationship between liquidity and bank profitability, Guru et al (2002) and Molyneux and Thornton (1992) report an opposite result.

3.4 Regulatory Capital Channel

• The regulatory capital arbitrage implies that banks securitize assets with relatively lower risk than those assets retained in their portfolio.
• Furthermore, Affinito and Tagliaferri (2010) show that banks mainly engage in securitization to reduce risk, improve liquidity, and improve capital ratios.
• Additionally, capital plays a pivotal role in determining bank profitability and is widely argued to have a positive impact on profitability.
• Some studies document negative impact of regulatory capital on bank profitability.
• Thus, based on the preceding discussion, the authors can formulate the fourth hypothesis as:.

3.5 Simultaneous Impact

• The preceding discussion shows that securitization is found to affect bank risk, cost of funding, liquidity, and regulatory capital, even though, the sign of this effect is not conclusive.
• Meanwhile, those factors affected by securitization are also determinants of bank profitability.
• The authors should also expect that the impacts of securitization and intermediate variables are simultaneous.
• Therefore, it can be argued that securitization transfers its effects to profitability simultaneously through a set of intermediate variables including bank risk, cost of funding, liquidity and regulatory capital.
• Thus, the authors can formulate their fifth hypothesis as:.

4.1 Econometric Specification

• The main hypothesis that the authors test is that bank risk, cost of funding, liquidity and regulatory.
• The impact of securitization on bank profitability and the four transmission channels: bank risk, cost of funding, liquidity and regulatory capital capital work as transmission channels between securitization and bank profitability.
• The authors start by investigating whether there is any potential effect resulting from the bank engagement in securitization activities on its profitability.
• Then the authors outline their model of testing and evaluating mediation effects in the securitization-profitability relationship.

4.1.1 Does Securitization Affect Bank Profitability?

• To empirically investigate the relationship between securitization and bank profitability of banks that engage in securitization activities, the authors estimate the following bank-specific fixed effects panel data model: PROF i,t = α i + !.
• The authors estimate the model with clustered standard errors at the bank level, which enables us to use within-bank variations to estimate the parameters of the relationship between securitization and profitability.
• The outstanding amount of securitization is used as the main explanatory variable for explaining the variation in bank profitability.
• In addition, some variables are included to control for the bank-specific characteristics, including loans to assets ratio, capital ratio, bank size, real GDP growth, trading assets ratio, loans to deposits, market share, and deposits to assets ratio.
• Variables are described in detail in section 4.2 below.

4.1.2 Identifying Individual Transmission Channels

• The direct relation between securitization and profitability can be expressed as follows: EQUATION Starting from a no mediation status is necessary to construct significance tests and to assess to what extent the direct effect of the independent variable is impaired by introducing a mediator into the relationship (Baron and Kenny, 1986) .
• Next, the authors introduce mediators into the securitization-profitability relationship.
• Finally, the authors account for both direct and indirect effects of securitization on bank profitability after introducing the mediator as follows: EQUATION First, SEM fits the model equations to data simultaneously, combining all the linear equations into one, using matrices and vectors (Cheong and MacKinnon, 2012) .
• Second, bootstrap procedures based on Maximum Likelihood (ML) estimation method can be applied to estimate the coefficients in a SEM, which provides more reliable and unbiased estimations for the indirect effect and enables us to infer more accurately about mediation (Cheong and MacKinnon, 2012) .
• Compared with regression-based approach, SEM is capable of handling complicated models that incorporate multiple mediators or those that use variables measured by multiple indicators (Cheong and MacKinnon, 2012) .

4.1.3 Identifying Simultaneous Transmission Channels

• The next step in testing the hypothesized mediation model is to construct and test a model that incorporates all the four proposed mediators at the same time.
• The aim here is to assess the direct and indirect effects of securitization on bank profitability in a more dynamic way that mimics the reality of this relationship.
• In addition, the complete mediation model helps to divide the indirect effect between mediators and estimate the percentage contribution of each one in transferring the effect from securitization to bank profitability (Iacobucci, 2008) .
• Also, the authors use bootstrap techniques to generate confidence intervals for the indirect effects to establish and classify mediation and to test for significance as outlined by Zhao et al. (2010) .

4.1.4 Significance of Transmission Channels

• The key to infer a mediated relationship between securitization and bank profitability through any of the previous proposed models is to test the significance of the indirect effects.
• In other words, z-test tests whether the mediated path (γ β) is statistically different from zero.
• This renders the test biased towards not rejecting the null hypothesis more often and consequently concluding no mediation.
• Therefore, bootstrapping techniques can be used to overcome this issue of the z-test (Zhao et al., 2010) .
• The proposed bootstrap test relies on the actual distribution of the indirect path coefficients (γ β) to construct confidence intervals for the indirect effect (Cheong and MacKinnon, 2012) .

4.1.5 Contribution of Transmission Channels

• The final step in analyzing the theoretical mediation model is computing the percentage contribution of each mediator in explaining the variation in bank profitability.
• This step is important to fully understand and visualize the complex relationship between securitization and bank profitability as represented in the four-mediator model.

4.2 Variables

• The main dependent variable in their paper is bank profitability.
• Following previous research, bank profitability is measured by either return on assets (ROA) or net interest margin (NIM) (Berger et al., 1995) .
• The authors use two measures for bank risk including the ratio of risk-weighted assets to total assets following Berger and Bouwman (2013) , and the ratio of charge-offs to total loans following Casu et al. (2013) .
• To measure the on-balance sheet liquidity of the bank, a widely-used measure is core liquidity ratio which is estimated as the ratio of cash to total assets, and the liquidity ratio (LIQ) which is estimated as the ratio of cash and securities to total assets.
• Finally, the authors use two measures for cost of funding including interest expense to total liabilities and interest expense to total deposits .

Exogenous Variables.

• These measures are lagged one period to allow the effects of securitization activities to be realized in bank profitability.
• Securitization is expected to liquidate current loans and provide the bank with an opportunity to grant new loans based on the new higher rates, in addition it provides other revenues arising from servicing fees (Casu et al., 2013) .
• Thus, the association between a bank's outstanding securitization and its profitability is expected to be positive.
• The authors also use a few control variables to account for the balance sheet heterogeneity among banks.
• The authors also include the loans to deposits ratio and the deposits total assets ratio to account for the stability of bank funding.

4.3.1 Data Sample

• The authors use data on US commercial banks including balance sheet information and securitization activities.
• This is necessitated by their proposed empirical model of the effects of securitization on bank profitability, which requires data on securitizers only.
• In addition, the authors obtain data on macroeconomic variables form the Federal Reserve Economic Data (FRED) database.
• To prevent the possibility of outliers driving the results, quarterly variables computed from the dataset are winsorized at the 1% level, that is, the smallest and largest 1% of the values of each variable are replaced with the closest value.
• The final dataset consists of 4842 bankquarters observations for 595 commercial banks.

4.3.2 Summary Statistics

• Table 1 provides summary statistics for the variables used to test the mediation model.
• This is in line with the fact that securitizers have additional sources of income from securitization activities such as servicing fees and trading revenue.
• Moreover, panel B provides statistics on cost of funding measures that shows higher disturbance in interest expense to total deposits ratio compared to interest expense to total liabilities ratio (standard deviation of 2.310 and 0.010 respectively).
• Turning to the regulatory capital measures, panel C shows that they are generally consistent with each other.
• Moreover, they tend to hold a relatively small amount of equity capital, 11% on average, which might reflect the fact that they have access to funds through the securitization market.

5 Results and Discussion

• This section presents the results of testing the empirical mediation model as specified in the methodology section above.
• The mediation models are estimated using SEM based on the Maximum Likelihood estimation method and a bootstrap procedure with 2000 iterations to construct a 95% confidence interval for the coefficients, direct, indirect, and total effects.
• On different dimensions of bank performance including profitability.
• The profitability dimension was found to be significantly affected by securitization activities.
• Similarly, Jiangli and Pritsker (2008), Lockwood et al. (1996) and Thomas (1999) suggest a positive impact of securitization on profitability using data on US commercial banks and bank holding companies.

5.2 Results of Individual Transmission Channels

• Having established the basic relationship between securitization and bank profitability, the authors can now move further to investigate the role that proposed mediators play in this relationship.
• Turning to the mediating role of bank risk, the analysis shows that both indirect and direct effects are significant (p < 0.01 for both), additionally the Sobel test is significant (z = 13.369, p < 0.01).
• Based on the criteria of Zhao et al. (2010) , it can be concluded that liquidity significantly mediates the relationship between securitization and bank profitability with a percentage of 2% (0.005/0.331).
• Based on the criteria of Zhao et al. (2010) , the results show the absence of any direct effect and a complete mediation can be concluded.

Table 3: Results of Analysing the Individual Transmission Channels

• This table provides the results of analysing the four transmission channels that mediate the relationship between securitization and bank profitability.
• The authors report standardized coefficients along with their 95% confidence intervals and p values as estimated by a bootstrap procedure based on 2000 iterations.
• While bank risk and cost of funding are found to positively affect profitability, previous studies suggest a negative effect for both.
• This contradiction can be justified in two ways.
• First, banks may intensively engage in securitization activities applying a generate-to-sell model and accepting to take more risk (see for example Bedendo and Bruno, 2012; Nijskens and Wagner, 2011) , but increase their income from non-interest income in the form of servicing fees (Casu et al., 2013) or trading activities (Minton et al., 2009) .

5.3 Results of Simultaneous Impact of Transmission Channels

• The individual mediation models were shown to be significant, and the proposed mediators are shown to be valid individually.
• The next step is to test the full hypothesized mediation model.
• The aim here is to assess the direct and indirect effects of securitization on the bank profitability in a more dynamic way similar to that in reality, in other words, incorporating all the four proposed mediators at the same time into the model.
• The analysis shows that only two out of nine paths are not significant at p < 0.05 level or better (the direct path and the path from securitization to liquidity that are significant only at p < 0.10).

Table 4: Results of Analysing the Simultaneous Impact of the Transmission Channels

• This table provides the results of simultaneously analysing the four transmission channels that mediate the relationship between securitization and bank profitability.
• The authors report standardized coefficients along with their 95% confidence intervals and p values as estimated by a bootstrap procedure based on 2000 iterations.

Table 5: Results of Analysing the Mediation Model without Direct Effect

• This table provides the results of a robustness check in which the authors simultaneously analyze the four transmission channels that mediate the relationship between securitization and bank profitability after removing direct effect of securitization.
• The authors report standardized coefficients along with their 95% confidence intervals and p values as estimated by a bootstrap procedure based on 2000 iterations.

5.4.1 Full Mediation Model

• The authors estimate the theoretical mediation model again after eliminating the direct effect of securitization.
• As shown in Table 5 , the fully mediated model fits the data appropriately.
• To compare the two models, the criteria suggested by (Bentler, 1990) The comparison shows that both models are equivalent on the overall statistics fit CFI (1.00).
• On the other hand, seven out of eight of the fully mediated model's paths were significant at the same level.
• To sum up, some minor differences exist between the two models for some criteria, but both models seem to equivalently and comparably explain the mediation model of the securitization-profitability relationship.

5.4.2 Adding Control Variables

• Furthermore, another test for the robustness of the results obtained using the original empirical model is to insert some control variables into the model and monitoring the change in the securitization-bank profitability relationship.
• Control variables are defined in Table B .1.
• The results of re-estimating the mediation model after including the control variables are presented in Table 6 .
• Based on these results, the authors can conclude some interesting findings.
• Fourth, the proportion of the indirect effect to the total effect increased to nearly 99.7% compared to 93% in the original model.

5.4.3 Using Alternative Measures

• Another robustness check is to repeat the analysis using alternative measures of securitization, bank profitability and mediators.
• Five different models are estimated based on the same methodology outlined by the empirical model.
• Specifically, model (2) uses interest to deposits ratio to measure the cost of funding, Tier 1 risk based capital (TIER1RBC) to measure regulatory capital, charge offs ratio to measure bank risk, and cash assets to total assets ratio to measure liquidity.
• While the results of models (1), ( 2) and ( 5) This table provides the results of a robustness check in which the authors simultaneously analyze the four transmission channels that mediate the relationship between securitization and bank profitability while including control variables.
• 35 support the existence of full mediation indicated by a significant indirect effect and an insignificant direct effect, model (4) shows a case of partial mediation with both direct and indirect effects being significant.

5.5 Discussion

• Overall, the results presented here provide a better understanding of the channels through which securitization affects bank profitability.
• Therefore, their findings provide some useful insights for banks.
• These decisions together would improve the design of securitization transactions and help banks to use securitization activities to boost their profitability while limiting the adverse effects on their soundness.
• One such initiative is the ECB's framework for simple, transparent, and standardized (STS) securitization (Mersch, 2017) .
• In addition, it requires banks to provide loanlevel information for ABSs if used as collateral in the Eurosystem's credit operations (ECB and BoE, 2014) .

6 Conclusion

• This paper contributes to the literature by exploring in depth the channels through which securitization impacts bank profitability.
• To this end, the authors use a mediation model to thoroughly investigate how securitization affects bank profitability.
• Also, the authors use a novel empirical framework based on structural equations modeling that simultaneously test the different relationships comprised in the proposed mediation model.
• There are some opportunities to extend the current analysis in different dimensions.
• The ratio of bank's total assets to the sum of all banks assets +/-GDPG GDP real growth.

Figures

Table 4: Results of Analysing the Simultaneous Impact of the Transmission Channels

Table 6: Results of Analysing the Mediation Model with Control Variables

Table 7: Results of Analysing the Mediation Model Using Alternative Variable Measures

Table 2: Results of Analysing the Impact of Securitization on Bank Profitability.

Table B.1: Variables Definition

Figure 1: The impact of securitization on bank profitability and the four transmission channels: bank risk, cost of funding, liquidity and regulatory capital

Table 5: Results of Analysing the Mediation Model without Direct Effect

Frequently asked questions

Q1. What have the authors contributed in "Disentangling the impact of securitization on bank profitability" ?

To this end, the authors analyze the role played by bank risk, cost of funding, liquidity and regulatory capital in explaining the relationship between securitization and bank profitability. The authors also show that bank risk, cost of funding, liquidity and regulatory capital individually and jointly act as transmission channels in the securitization-profitability relationship. 

Q2. What future works have the authors mentioned in the paper "Disentangling the impact of securitization on bank profitability" ?

There are some opportunities to extend the current analysis in different dimensions. Additionally, further analysis can be done to investigate the impact of different types of securitization, including ABS or MBS, on bank profitability. 

Q3. What is the main explanatory variable for the variation in bank profitability?

The outstanding amount of securitization is used as the main explanatory variable for explaining the variation in bank profitability. 

Q4. What is the expected effect of the size of a bank on its profitability?

The bank size is expected to positively affect its profitability due to the economies of scale or the ability of large banks to lend more. 

Q5. What does Loutskina argue that reduces the cost of funding?

Loutskina (2011) argues that the availability of securitization as an internal source of funds reduces the sensitivity of the bank cost of funding to the availability of other external sources of funds such as traditional liquid funds and deposits. 

Q6. What is the main reason why banks securitize?

Jones (2000) finds that banks securitize, among other reasons, to diversify funding sources, reduce the costs of external financing through debt and deposits, and accordingly to reduce the overall cost of funding. 

Q7. What are the main components of the securitization relationship?

Their results show that bank risk, cost of funding, liquidity and regulatory capital work as transmission channels in the securitizationprofitability relationship. 

Q8. What is the origin of mortgage loans in the US?

In the US, securitization origins go back to the early 1970s, when Government National Mortgage Association (Ginnie Mae) started to sell mortgage loans (Ibanez and Scheicher, 2012). 

Q9. What are the variables used to measure the liquidity of a bank?

To measure the on-balance sheet liquidity of the bank, a widely-used measure is core liquidity ratio (CORELIQ) which is estimated as the ratio of cash to total assets, and the liquidity ratio (LIQ) which is estimated as the ratio of cash and securities to total assets. 

Q10. How do the authors estimate the relationship between securitization and profitability?

The authors estimate the model with clustered standard errors at the bank level, which enables us to use within-bank variations to estimate the parameters of the relationship between securitization and profitability. 

Q11. What is the empirical framework used to test the different relationships?

their empirical framework is based on a Structural Equations Modeling (SEM) approach that can simultaneously test the different relationships comprised in the proposed empirical model. 

Q12. What was the perception of the US banking system until mid-2007?

until mid-2007 it was widely perceived that the US banking system was sound and performing well, particularly because banks capital holdings and profitability appeared to be high and at record levels. 

Q13. What did the increase in foreclosures and defaults in mortgages mean for the US banking industry?

The increased number of foreclosures and defaults in mortgages led to a decline in the value of securitized assets and reduced investors’ appetite for such securities and accordingly problems within the US banking industry (Gerardi et al., 2008). 

Q14. What is the average deviation of the risk-weighted assets to total assets ratio?

With respect to bank risk measures, panel D shows that risk-weighted assets to total assets ratio has a high standard deviation of 0.270 compared to 0.021 of the charge-offs ratio. 

Q15. What are the main channels through which securitization affects bank profitability?

In so doing, the authors argue that the impact of securitization on bank profitability is transmitted through four main channels, namely bank risk, cost of funding, liquidity and regulatory capital. 

Q16. What is the effect of securitization on bank risk?

On the other hand, securitization may increase bank risk due to the increase in risk taking and recourse or other seller-provided credit enhancements (Higgins and Mason, 2004). 

Q17. How does the model estimate the percentage of indirect effect?

In this regard, the authors follow the method applied by Iacobucci (2008) to estimate the percentage of indirect effect using the estimated indirect and direct path coefficients as follows:34 = 678 9:7678 9:7 1 6; (8)where km is the indirect effect that is transmitted through mediator m as a percentage of total effect. 

Q18. What is the next step in testing the hypothesized mediation model?

The next step in testing the hypothesized mediation model is to construct and test a model that incorporates all the four proposed mediators at the same time. 

Q19. What is the final step in analyzing the theoretical mediation model?

The final step in analyzing the theoretical mediation model is computing the percentage contribution of each mediator in explaining the variation in bank profitability.