Three essays on credit supply

Pinheiro, Jorge (2023) Three essays on credit supply. PhD thesis, University of Glasgow.

Abstract

This thesis consists of three independent essays on credit supply, each addressing different components, including the different impact of credit supply shocks financed through different supply channels, how different credit constraints impact debt structure and productivity, and how it affects their individual and collective exposure over time.

Chapter 1: Its conceptual appeal has made the Conditional Value at Risk (CoVaR) one of the most influential systemic risk indicators. Despite its popularity, an outstanding methodological challenge may hamper the CoVaRs’ accuracy in measuring the time-series dimension of systemic risk. The dynamics of the CoVaR are entirely due to the behaviour of the state variables and therefore without their inclusion, the CoVaR would be constant over time. The key contribution of this chapter is to relax the assumption of time-invariant tail dependence between the financial system and each institution’s losses, by allowing the estimated parameters of the model to change over time, in addition to changing over quantiles and different financial institutions. We find that the dynamic component that we introduce does not affect the estimations for the risk of individual financial institutions, but it largely affects estimations of systemic risk which exhibits more procyclicality than the one implied by the standard CoVaR. As expected, larger financial institutions have a higher effect on systemic risk, although they are also shown to be individually more robust. When adding balance sheet data, it introduces additional volatility into our model relative to the standard one. In terms of forecasting, the results depend on the horizon used or the variables included. There is no clear outperformance between either model when we add the balance sheet data, or in the short term (less than 12 weeks). However, our model outperforms the standard one for medium (between 15 and 25 weeks) to long term horizons (between 30 and 40 weeks).

Chapter 2: We seek to evaluate the impact of the different segments within the lending sector to the private non-financial sector can have on subsequent GDP growth. We isolate the bank lending channel as one of the main components, and group the remaining ones into a second segment which we classify as market based finance (MBF). We also include the 2 different segments of the borrowing sector, household debt and non-financial firm debt, to compare with the results obtained by the standard model. We debate the main source of these effects, and focus on either credit demand or credit supply shocks, in addition to other alternatives. We find that a rise in bank credit and/or household debt to GDP ratio lowers subsequent GDP growth. The predictive power is large in magnitude and robust across time and space. The bank credit booms and household debt booms are connected to lower interest rate spread environments, as well as periods with better financial conditions. And although the overall impact on subsequent GDP growth is negative, we found contrasting evidence when using the Financial Conditions Index (FCI) as an instrument. This would point to the potential different effects that bank credit and household debt could have on future economic growth (good booms vs bad booms), depending on the underlying cause of the boom. The results and the evidence that we found are more consistent with models where the fundamental source of the changes in household debt or bank credit lie in changes in the credit supply (credit supply shocks), rather than credit demand or other possibilities. This would likely be connected to incorrect expectations formation by lenders and investors (what many authors classify as “credit market sentiment” in the literature), which is an important element in explaining shifts in credit supply. Although credit demand shocks could play an important role in prolonging or amplifying the effects of the booms, it is unlikely that they are the source, as it would lead to results that conflict with empirical evidence. Finally, we find some differences in terms of statistical significance and magnitude in the different scenarios, where the bank credit shows more robustness to different specifications than the household debt. This would imply that there is a significance of the bank credit that goes well beyond the household debt. It would also mean that the main component that generates the boom bust cycle in GDP would be the bank credit, independent of its destination, rather than household debt, independent of its financing.

Chapter 3: We construct a dataset at the firm-year level by merging the syndicated loan data, provided by Refinitiv LPC DealScan ("DealScan"), with the firm level data, provided by Center for Research in Security Prices (CRSP)/Compustat Merged Database ("CCM"). We conduct an analysis on firms subjected to different covenants, and find that firms with earnings-based constraints have lower levels of TFP (Total Factor Productivity), and short-term debt, when compared to firms with asset-based constraints. The data also shows that this is connected to an additional negative impact that short-term debt has on the productivity for the firms with earnings based constraints, which does not verify in the firms with asset-based constraints. Both these characteristics are robust to the use of 3 different TFP estimation methods, different subsamples, and additional controls, including age and size of the firm. Thus, we consider a quantitative dynamic stochastic partial equilibrium model, with three main types of firms, distinguished by their constraints, which explores the impact of short-term and long term borrowing on firm’s balance sheets, on the different variables. We construct replications for this theoretical model, and assess the how well it fits our actual data. Our findings show that constraints exert an impact on short-term borrowing, but not on the remaining variables. More specifically, firms that face an earnings-based constraint show lower levels of short-term borrowing, compared with firms that are either unconstrained, or asset-based constraint. The adjustment is made through lower dividend distribution, as can be seen by the lower values of the value function. They also point to the impact being larger for firms with lower productivity shocks, which is in accordance withour empirical findings. Even though that our data shows
differences in some of this variables (for example, on long-term debt), these were not robust to some of the controls, including the size of the firm.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: H Social Sciences > HB Economic Theory; H Social Sciences > HG Finance
• Colleges/Schools: College of Social Sciences > Adam Smith Business School > Economics
• Supervisor's Name: Tsoukalas, Professor John, Nolan, Professor Charles and Tsafo, Dr. Yannis
• Date of Award: 2023
• Depositing User: Theses Team
• Unique ID: glathesis:2023-83814
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 13 Sep 2023 10:07
• Last Modified: 13 Sep 2023 10:09
• Thesis DOI: 10.5525/gla.thesis.83814
• URI: https://theses.gla.ac.uk/id/eprint/83814