SLO Study

Introduction

In this chapter we present an application to show how the material we have covered so far in this book can be used to put together signals about the market. In particular, a signal that may not be widely followed and could give us an edge. Here we draw on academic work from Chava, Gallmeyer and Park (2015) who showed that a survey of bank loan officers conducted by the Fed has robust predictive power for market returns.1

The Federal Reserve Board’s Senior Loan Officer Opinion Survey on Bank Lending Practices (the “SLO”) is a poll of the major U.S. banks about credit conditions. According to the Fed, up to 80 domestic banks and 24 U.S. branches of foreign banks participate in the survey. Lown and Morgan (2006) estimate these banks account for “about 60% of all loans by U.S. banks and about 70% of all U.S. bank business loans.”

The SLO is a quarterly survey, timed around four meetings of the Federal Open Market Committee. Questions relate to the terms of the banks’ lending and the state of business and household demand for loans; in addition, the survey includes a handful of questions on current topics. The actual results of the latest survey can be viewed at https://www.federalreserve.gov/data/sloos.htm

As an example, the following is the first question of the current wave of the SLO:

Over the past three months, how have your bank’s credit standards for approving applications for C&I loans or credit lines – other than those to be used to finance mergers and acquisitions – to large and middle-market firms and to small firms changed?

The lending officers are provided the following choices:

  • Tightened considerably
  • Tightened somewhat
  • Remained basically unchanged
  • Eased somewhat
  • Eased considerably

The responses are tallied, organized, and presented only in aggregate form. The market’s preferred way to look at this report is by the proportion of banks that are tightening their lending standards versus the ones that are easing. This survey can serve as an early signal on the macroeconomy: if lending standards tighten, it becomes more difficult for companies to obtain funding for new investments or roll over previous debt, which in turn may impact their cash flows and returns. Chava, Gallmeyer and Park (2015) found that tighter standards can predict lower market returns the following quarter.

Let’s get to work by pulling the SLO data and visualizing it.

Importing the Data

The SLO survey was first conducted back in 1967 but its modern form starts in the second quarter of 1990. The survey has evolved over time and some sections that we may consider crucial today were not part of the survey at the beginning. For instance, questions about lending standards of credit card loans began in 1996, information about standards of subprime mortgage loans until 2007, and questions about auto loans start in 2011. The longest time series, the one most commentators refer to when they talk about the SLO, is about lending standards for commercial and industrial (C&I) loans, which are separated into small firms and large and middle-market firms.

Let’s pull all of these series from FRED. The SLO series in FRED represent the net percentage of respondents that are tightening, this is the percentage of respondents that tightened standards considerably or somewhat minus the percentage that eased somewhat or considerably. When the series is positive, then more banks are tightening, and when it is negative then more banks are easing.

SLO = (
    pdr.get_data_fred(['DRTSCILM', 'DRTSCIS', 'DRTSCLCC', 'DRTSSP', 'STDSAUTO'],
                    start="1990-04-01")
        .reset_index()
        .rename(columns={'DATE': 'Date',
                        'DRTSCILM': 'Large_and_Mid',
                        'DRTSCIS': 'Small',
                        'DRTSCLCC': 'Credit_cards',
                        'DRTSSP': 'Subprime_mortgage',
                        'STDSAUTO': 'Auto_loans'})
        .assign(qtr_year = lambda df: df['Date'].dt.to_period('Q'),
                Large_and_Mid = lambda df: df['Large_and_Mid']/100,
                Small = lambda df: df['Small']/100,
                Credit_cards = lambda df: df['Credit_cards']/100,
                Subprime_mortgage = lambda df: df['Subprime_mortgage']/100,
                Auto_loans= lambda df: df['Auto_loans']/100)
)

SLO.head()
        Date  Large_and_Mid  Small  ...  Subprime_mortgage  Auto_loans  \
0 1990-04-01          0.544  0.527  ...             NaN            NaN   
1 1990-07-01          0.467  0.339  ...             NaN            NaN   
2 1990-10-01          0.542  0.407  ...             NaN            NaN   
3 1991-01-01          0.386  0.316  ...             NaN            NaN   
4 1991-04-01          0.200  0.069  ...             NaN            NaN   

   qtr_year  
0    1990Q2  
1    1990Q3  
2    1990Q4  
3    1991Q1  
4    1991Q2  

[5 rows x 7 columns]

Information is lacking for credit cards, subprime mortgages, and auto loans at the start of our sample, but that’s fine, let’s visualize all series together. We will first melt the dataframe so we have all categories stacked or long.

SLO_long = pd.melt(SLO.drop(columns='qtr_year'),
                id_vars = 'Date',
                var_name = 'Series',
                value_name = 'Value')

Let’s plot the series and include recession shades.

SLO and Market Returns

Conclusion

Footnotes

  1. Chava, S., Gallmeyer, M., & Park, H. (2015). “Credit conditions and stock return predictability”. Journal of Monetary Economics, 74, 117-132. The working paper version can be found at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1571958↩︎