Evidence of Forward-Looking Loan Loss Provisioning with Credit Market Information

Abstract

Our paper makes a fundamental contribution by studying loan loss provisioning over the credit cycle as three distinct phases. Looking at the three distinct phases of the financial crisis – the pre-crisis period, crisis period, and post-crisis period – is important as loan loss provisioning is driven by different factors in each, in part due to extensive shifts in (or in the application of) regulatory rule. Controlling for credit market information using data from the Senior Loan Officer Opinion Surveys (SLOOS) we extend the work of previous studies of forward-looking loan loss provisions using the delayed expected loss recognition approach. We contribute to the growing literature on forward-looking loan loss provisioning and early in the cycle loss recognition by incorporating a broader range of available credit information and explicitly controlling for structural breaks in the sample corresponding to the financial crisis.

Appendix

1.1 Primer on Senior Loans Officer Opinion Survey (SLOOS)

The Federal Reserve’ Senior Loans Officer Opinion Survey on Bank Lending Practices is a qualitative survey that includes close to twenty core questions related to supply and demand for various categories of credit. Depending on the prevailing economic and financial conditions, the survey may include additional ad hoc questions specific to problems and trends in the credit markets. The survey is usually conducted four times per year, keeping in view the schedule of the meetings of the Federal Open Market Committee (FOMC). As a result, the SLOOS survey is a quarterly survey that can take place at different points within a quarter. Even though the surveys are routinely conducted four times a year, the Federal Reserve Board has the authority to conduct up to six surveys in a year. The extra surveys are typically reserved for volatile times when the financial and credit markets are unstable. For our study, we do not use the information from extra surveys. It is important to note that the SLOOS survey includes reported changes in lending standards and loan demand over the three months preceding the date the survey was distributed. So when merging it with other quarterly data sources one need to adjust for this fact. In our case, we matched the quarter of the Call Report data with the quarter of SLOOS responses. For example, the January SLOOS corresponds to the SLOOS responses reported over the period October to December of the previous year. Accordingly, we merge the January SLOOS with the fourth quarter Call report data.

The SLOOS sample size is modest as it includes about roughly 60 large domestic banks (recently it has been expanded to 80). All of these banks are headquartered in one of the twelve Federal Reserve Districts, with a minimum of two and a maximum of twelve from each district. Given the increasing concentration of banking sector assets among large banks, the survey is intentionally weighted towards the large banks because doing so will allow it to capture and monitor a significant fraction of the total loans outstanding within the banking system. It also permits responses for each of the loan categories covered by the survey, since big banks are likely to be lending in all main loan categories.

The participation in the survey is voluntary but the response rate is almost 100 %, meaning that banks that are requested to participate almost always do. Furthermore, even though survey participants have an option not to respond to any specific question they almost always responds. The main reason the banks drop out of the panel is due to acquisition by another SLOOS bank in the panel.

One very important aspect of the SLOOS is that their identity and individual responses are kept confidential and specifically not shared with the supervision and regulation staff at the Federal Reserve System. The primary reason behind this is to insure accurate and honest responses from the banks without worrying the impact their responses would have on the actions of their regulators. And so in reporting the summary statistics and other results we are very being extremely careful and so do not report minimum, maximum and other statistics that may compromise the identity and responses of the SLOOS participants.

For a detailed and complete description of the SLOOS such as panel selection criteria, methodology, timing of the surveys, exact questions and their wordings, loan categories covered please refer to http://www.federalreserve.gov/boarddocs/SnLoanSurvey and Bassett et al. (2012).

1.2 Construction of the diffusion indexes: lending standards and loan demand (same as documented in Bassett et al. (2012))

In SLOOS, the loan officers are asked whether they have changed lending standards since the quarter before for the following loan categories: commercial and industrial, commercial real estate, residential mortgages to buy homes, home equity lines of credit, and consumer loans (auto loans, credit cards, and consumer loans other than credit cards or auto loans). In our study we will instead work with first three loan categories: commercial and industrial, commercial real estate, residential mortgages to buy homes. For each of these loan categories, the loan officers are also asked about their perception of the changes in loan demand.

A typical question about changes in standards looks like the following (consider for C&I loan category):

“Over the past three months, how have your bank’s credit standards for approving loan applications for C&I loans or credit lines changed?”

The multiple-choice answers:

1) Eased considerably, 2) eased somewhat, 3) about unchanged, 4) tightened somewhat, 5) tightened considerably.

Similarly, a typical question about changes in demand looks like the following (consider for C&I loan category):

“Over the past three months, how has the demand for C&I loans or credit lines at your bank changed?”

The possible answers:

1) increased considerably, 2) increased somewhat, 3) about unchanged, 4) decreased somewhat, 5) decreased considerably.

Given in the past, loan officers have hardly ever characterized changes in either standards or demands as ‘considerably’, we therefore simplify our analysis by recoding the reported responses into three categories and accordingly create lending and demand categorical variables respectively.

Step 1: Creating categorical variables

The lending categorical variable, $\overset{-}{{\boldsymbol{S}}_{\boldsymbol{it}}}$ [c]

$\overset{-}{{\boldsymbol{S}}_{\boldsymbol{it}}}$ [c]

−1 if bank i reported easing standards on loan category c in quarter t

0 if bank i reported no change in standards on loan category c in quarter t

+1 if bank i reported tightening standards on loan category c in quarter t

And similarly the demand categorical variable, $\overset{-}{{\boldsymbol{D}}_{\boldsymbol{it}}}$ [c]

$\overset{-}{{\boldsymbol{D}}_{\boldsymbol{it}}}$ [c]

−1 if bank i reported decreased demand for loan category c in quarter t

0 if bank i reported no change in demand for loan category c in quarter t

+1 if bank i reported increased demand for loan category c in quarter t

Step 2: Constructing a diffusion index for changes in lending standards and one for changes in loan demand

Next we construct a composite or diffusion index of changes in lending standards and loan demand for each bank in our panel as weighted averages:

$$ \varDelta \kern0.5em \boldsymbol{Standard}{\boldsymbol{s}}_{\boldsymbol{it}}={\displaystyle {\sum}_{\boldsymbol{c}}{\boldsymbol{w}}_{\boldsymbol{it}}\left[\boldsymbol{c}\right]}\times \cdot \overline{{\boldsymbol{S}}_{\boldsymbol{it}}}\cdot \left[\mathbf{c}\right] $$

$$ \varDelta \kern0.5em \boldsymbol{Deman}{\boldsymbol{d}}_{\boldsymbol{it}}={\displaystyle {\sum}_{\boldsymbol{c}}{\boldsymbol{w}}_{\boldsymbol{it}}\left[\boldsymbol{c}\right]}\times \cdot \overline{{\boldsymbol{D}}_{\boldsymbol{it}}}\cdot \left[\mathbf{c}\right] $$

where 0 ≤ w _it [c] ≤ 1 represents the fraction of bank i’s core loan portfolio that consists of three loan categories in category c, as reported on bank i’s Call Report in quarter t.

∆Standards _it and ∆Demand _it takes on continuous values between −1 and +1.

Interpretation:

∆Standards _it represents the net fraction of loans on bank i’s balance sheet that were in categories for which bank reported changing lending standards over the survey period.

∆Demand _it represents the net fraction of loans on bank i’s balance sheet that were in categories for which bank (as reported) experienced a change in demand over the survey period.

Step 3: (Optional) Constructing an aggregate diffusion index for changes in lending standards and one for changes in loan demand (not used in the regressions)

In the previous step we constructed bank specific composite indexes, which can be aggregated across banks to come up with an aggregate composite or diffusion indexes:

$$ \varDelta \kern0.5em \boldsymbol{Standard}{\boldsymbol{s}}_{\boldsymbol{t}}={\displaystyle {\sum}_{\boldsymbol{i}}{\boldsymbol{w}}_{\boldsymbol{i}\boldsymbol{t}}}\times \varDelta \kern0.5em \boldsymbol{Standard}{\boldsymbol{s}}_{\boldsymbol{i}\boldsymbol{t}} $$

$$ \varDelta \kern0.5em \boldsymbol{Deman}{\boldsymbol{d}}_{\boldsymbol{t}}={\displaystyle {\sum}_{\boldsymbol{i}}{\boldsymbol{w}}_{\boldsymbol{i}\boldsymbol{t}}}\times \varDelta \kern0.5em \boldsymbol{Deman}{\boldsymbol{d}}_{\boldsymbol{i}\boldsymbol{t}} $$

where 0 ≤ w _it  ≤ 1 represents the fraction of total core loans on SLOOS respondents’ balance sheets that are held by bank i in quarter t.

∆Standards _t and ∆Demand _t takes on continuous values between − 1 and + 1.

These indices summarize the economy wide changes in credit supply and demand.

Figure 1b plots both the aggregate lending standards and loan demand alongside the real GDP growth.

1.3 Bank mergers

As mentioned earlier, the primary cause of attrition in the SLOOS is due to bank mergers. Accordingly we adjust for the bank mergers in our sample as follows (see also English and Nelson 1998):

When banks merge they are two possible accounting methods that are used to handle the merger. One of those accounting approach is called ‘purchase accounting’. Under this approach, the balance sheet items of the acquired bank are combined together and reported in the quarter of the merger, but the year-to-date flow of income and expense of the acquired bank as of the date of merger is not reported by the acquiring institution after the merger. Whereas in the second accounting approach called ‘pooling of interest accounting’ both balance sheets and income statements of the merging banks are combined and reported as of the date of the merger. Luckily the sample period we are working with identifies the accounting method used for bank mergers. Specifically we use the bank merger data files that are publicly available from the Federal Reserve Bank of Chicago’s website (http://www.chicagofed.org/webpages/publications/financial_institution_reports/merger_data.cfm) to identify bank mergers in our sample. We keep the bank mergers that used pooling of interest accounting but discard those that used purchase accounting. That is we drop those observations corresponding to the quarter in which the merger took place and the accounting method used was purchase accounting. The observations dropped amounted to 6 % of the sample.

1.4 Call report data description

VARIABLE DESCRIPTION	CALL REPORT ITEMS/DATA SOURCE	MNEMONICS
C&I Loans
COMMERCIAL AND INDUSTRIAL LOANS	RCFD1766	TBL
SMALL FIRM LENDING	RCON5571 + RCON5573 + RCON5575	SBL
SMALL FIRM LENDING WEIGHT	(RCON5571 + RCON5573 + RCON5575) /(RCFD1766)	SML_RATIO
LARGE FIRM FIRM WEIGHT	1- {SML_RATIO}	LRG_RATIO
C&I LOANS 90 DAYS OR MORE PD & NONACCRUAL	RCFD 1607 + RCFD 1608	BAD_CI_LNS_90
C&I LOANS 90 DAYS OR MORE PD & NONACCRUAL FOR SMALL FIRMS	SML_RATIO* (RCFD 1607 + RCFD 1608 )	BAD_CI_LNS_90_SML
C&I LOANS 90 DAYS OR MORE PD & NONACCRUAL FOR LARGE MID SIZED FIRMS	LRG_RATIO * (RCFD 1607 + RCFD 1608)	BAD_CI_LNS_90_LRG
CHARGE-OFFS ON C&I LOANS	RIAD4638	CHGOFF_CI
RECOVERIES ON C&I LOANS	RIAD4608	RECOV_CI
NET CHARGEOFFS ON C&I LOANS	RIAD4638 - RIAD4608	NET_CHGOFF_CI
NET CHARGEOFFS ON C&I LOANS FOR SMALL FIRMS	SML_RATIO * (RIAD4638 - RIAD4608)	NET_CHGOFF_CI_SML
NET CHARGEOFFS ON C&I LOANS FOR LARGE AND MID SIZED FIRMS	LRG_RATIO * ((RIAD4638 - RIAD4608)	NET_CHGOFF_CI_LRG
CRE Loans
LOANS SECURED BY 1–4 FAMILY RESIDENTIAL CONSTRUCTION	RCONF158
LOANS SECURED BY OTHER CONSTRUCTION LOANS AND ALL LAND DEVELOPMENT AND OTHER LAND LOANS	RCONF159
REAL ESTATE LOANS SECURED BY MULTI-FAMILY (5 OR MORE) RESIDENTIAL PROPERTIES	RCON1460
LOANS SECURED BY OWNER OCCUPIED NONFARM NONRESIDENTIAL PROPERTIES	RCONF160
LOANS SECURED BY OTHER NONFARM NONRESIDENTIAL PROPERTIES.	RCONF161
TOTAL CRE LOANS	RCONF158 + RCONF159 + RCON 1460 + RCONF160 + RCONF161	TOT_CRE
LOANS SECURED BY 1–4 FAMILY RESIDENTIAL CONSTRUCTION 30 DAYS PD + LOANS SECURED BY OTHER CONSTRUCT LOANS & ALL LAND DEVT AND OTHER LAND LOANS 90 DAYS PD	RCONF174 + RCONF175
NONACCRUAL	RCONF176 + RCONF177
REAL ESTATE LOANS SECURED BY MULTI-FAMILY (5 OR MORE) RESIDENTIAL PROPERTIES 90 DAYS PD	RCON3500
NONACCRUAL	RCON3501
LOANS SECURED BY OWNER OCCUPIED NONFARM NONRESIDENTIAL PROPERTIES 90 DAYS PD & NONACCRUAL + LOANS SECURED BY OTHER NONFARM NONRESIDENTIAL PROPERTIES 90 DAYS PD	RCONF180 + RCONF181
NONACCRUAL	RCONF182 + RCONF183
TOTAL BAD CRE LOANS 90 DAYS PD & NONACCRUAL	RCONF174 + RCONF175 + RCONF176 + RCONF177 + RCON3500 + RCON3501 + RCONF180 + RCONF181 + RCONF182 + RCONF183	BAD_CRE_LNS_90
LOANS SECURED BY 1–4 FAMILY RESIDENTIAL CONSTRUCTION & OTHER CHARGEOFFS	RIADC891 + RIADC893
LOANS SECURED BY 1–4 FAMILY RESIDENTIAL CONSTRUCTION RECOVERY	RIADC892 + RIADC894
LOANS SECURED BY 1–4 FAMILY RESIDENTIAL CONSTRUCTION & OTHER NET CHARGEOFFS	(RIADC891 + RIADC893) - (RIADC892 + RIADC894)
REAL ESTATE LOANS SECURED BY MULTI-FAMILY (5 OR MORE) RESIDENTIAL PROPERTIES CHARGEOFFS	RIAD3588
REAL ESTATE LOANS SECURED BY MULTI-FAMILY (5 OR MORE) RESIDENTIAL PROPERTIES RECOVERY	RIAD3589
REAL ESTATE LOANS SECURED BY MULTI-FAMILY (5 OR MORE) RESIDENTIAL PROPERTIES CHARGEOFFS	RIAD3588 - RIAD3589
LOANS SECURED BY OWNER OCCUPIED NONFARM NONRESIDENTIAL PROPERTIES CHARGEOFFS	RIADC895 + RIADC897
LOANS SECURED BY OWNER OCCUPIED NONFARM NONRESIDENTIAL PROPERTIES RECOVERY	RIADC896 + RIADC898
LOANS SECURED BY OWNER OCCUPIED NONFARM NONRESIDENTIAL PROPERTIES NET CHARGEOFFS	(RIADC895 + RIADC897) - (RIADC896 + RIADC898)
TOTAL NET CHARGEOFFS ON CRE LOANS	(RIADC891 + RIADC893) - (RIADC892 + RIADC894) + RIAD3588 - RIAD3589 + (RIADC895 + RIADC897) - (RIADC896 + RIADC898)	NET_CHGOFF_CRE
Residential Loans
REVOLVING, OPEN-END LOANS SECURED BY 1–4 FAMILY RESIDENTIAL PROPERTIES AND EXTENDED UNDER LINES OF CREDIT	RCON1797
ALL OTHER LOANS SECURED BY 1–4 FAMILY RESIDENTIAL PROPERTIES: SECURED BY FIRST LIENS	RCON5367
ALL OTHER LOANS SECURED BY 1–4 FAMILY RESIDENTIAL PROPERTIES: SECURED BY JUNIOR LIENS	RCON5368
TOTAL RESIDENTIAL LOANS	RCON1797 + RCON5367 + RCON5368	TOT_RESI
REVOLVING, OPEN-END LOANS SECURED BY 1–4 FAMILY RESIDENTIAL PROPERTIES AND EXTENDED UNDER LINES OF CREDIT 90 DAYS PD + ALL OTHER CLOSED END LOANS	RCON5399 + RCONC237 + RCONC239
NON ACCRUAL	RCON5400 + RCONC229 + RCONC230
TOTAL RESIDENTIAL LOANS 90 DAYS PD & NONACCRUAL	(RCON5399 + RCONC237 + RCONC239) + (RCON5400 + RCONC229 + RCONC230)	BAD_RESI_LNS_90
CHARGE-OFFS ON REVOLVING, OPEN-END LOANS SECURED BY 1–4 FAMILY RESIDENTIAL PROPERTIES AND EXTENDED UNDER LINES OF CREDIT & CLOSED END CREDIT	RIAD5411 + RIADC234 + RIADC235
RECOVERIES ON REVOLVING, OPEN-END LOANS SECURED BY 1–4 FAMILY RESIDENTIAL PROPERTIES AND EXTENDED UNDER LINES OF CREDIT & CLOSED END CREDIT	RIAD5412 + RIADC217 + RIADC218
TOTAL NET CHARGEOFFS ON RESIDENTIAL LOANS	(RIAD5411 + RIADC234 + RIADC235) - (RIAD5412 + RIADC217 + RIADC218)	NET_CHGOFF_RESI
Other
TOTAL EQUITY CAPITAL	RCFD3210	EQTY_CAP
TOTAL NONINTEREST INCOME	RIAD4079	NONINT_INC
INTEREST AND FEE INCOME ON LOANS, TOTAL	RIAD4010	LOAN_INT
TOTAL NONINTEREST EXPENSE	RIAD4093	NONINT_EXP
TOTAL INTEREST INCOME	RIAD4107	INT_INCOME
TOTAL INTEREST EXPENSE	RIAD4703	INT_EXP
PROVISION FOR LOAN AND LEASE LOSSES	RIAD4230	PLL
TOTAL ASSETS	RCFD2170	TOT_ASSETS
ALLOWANCE FOR LOAN AND LEASE LOSSES	RIAD3123	ALLL
EARNINGS BEFORE TAXES AND PROVISIONS	RIAD4301 + RIAD4230	EBTP