Labor Supply as a Buffer: The Implication of Credit Constraints in the US

Abstract

The credit constraint, an example of an incomplete credit market, provides an incentive to intensify the extensive and intensive margins related to labor force participation and work hours, respectively. This study uses the cross-section data from the Survey of Consumer Finance (SCF) and analyzes the impact of credit constraints on labor supply decisions, time to search for employment, and work hours. The empirical findings using the IV-probit and 2SLS models suggest that credit constraints and their various measures encourage households to increase both labor force participation and work hours to offset the negative impact of financial constraints. The intensity of working hours increases when we introduce both the alternate form of credit constraint and various age bands. Credit-constrained individuals effectively search for jobs and are most likely to accept employment in a short period, but their job search process takes more time than non-constrained individuals.

1. Introduction

The pioneering work of Deaton (1991) and Carroll (1992) analyzes household savings and consumption decisions with exogenous labor supply. In their life-cycle model, liquidity constraints can affect household consumption because consumers do not have the flexibility to make decisions about labor supply (Bui & Ume, 2020). However, recent developments in the literature, including the work of Low (2005) and Attanasio et al. (2005), explicitly consider the importance of labor supply decisions as a “buffer” and a margin of adjustment in the presence of borrowing constraints. In this modern literature, the individual decision to provide the labor supply is not only affected by frictions in the labor market, but financial frictions also play a significant role in the allocation of working hours, time to get reemployment, and labor force participation. Hence, financial shocks and credit imperfections have become the focus of attention to understand the response of labor supply and consumer spending (Bartscher, 2023; Blundell, 2006; Kumar & Liang, 2024). Credit constraints, an example of credit imperfection or an incomplete credit market, impact not only consumption but also labor supply decisions and the choice of work hours.

There are only a handful of papers that analyze liquidity constraints and their impact on the labor market. In addition, they have completely ignored the credit market, its various measures, and their impact on labor force participation and labor supply hours. We fill this gap in the literature by exploiting unique data, the Survey of Consumer Finance (SCF), a cross-sectional dataset collected by the Federal Reserve Board. Additionally, we also account for hidden measures of credit constraints, such as discouraged borrowers, and analyze their impacts on the labor market. In general, we answer a few research questions: (i) What is the impact of being turned down for a credit application, a measure of credit constraint, on labor force participation and working hours? (ii) How do rejected borrowers behave in the labor search process, and how long does it take them to find reemployment? (iii) How do different measures of credit constraint impact the household’s decision to respond to the labor supply? (iv) How do both credit constraints and hourly wage rates impact the household’s decision to supply labor? (v) Does the credit constraint affect the outcome of short-term job searches, a measure of the time it takes to find reemployment?

Financial frictions weaken individuals’ ability to borrow from the credit market, resulting in a myopic reaction due to liquidity constraints. This further creates asymmetric consumption between households, encouraging them to explore more options to join the workforce, reevaluate their labor–leisure trade-off, and increase their work hours (Garcia et al., 1997). Increasing work hours ultimately neutralizes the impact of credit constraints and helps people overcome their inability to borrow (Rossi & Trucchi, 2016). Furthermore, credit restrictions result in intra-household risk-sharing behavior and encourage the exploration of more options to offset the impact of financial difficulties on family members (Yue et al., 2021). Hence, imperfections in the credit market or limited access to credit act as a buffer to labor supply. From an employer’s perspective, a lack of capital and unavailability of family wealth have caused small firms to experience more credit constraints than large firms. Moreover, high risks in businesses make them more vulnerable to existing financing conditions and reduce their ability to obtain more credit from the market (Berger & Udell, 2006; Cowan et al., 2015; Stefania et al., 2022). Thus, credit constraints limit their financial portfolio, which restricts not only the productivity of firms but also discourages the job creation process. Higher firing costs can harm their expenses and output, particularly in the presence of financial restrictions, forcing firms to generate short-term liquidity (Bennett, 2011).

In addition, significant evidence indicates that employers offer additional work hours (Benito & Saleheen, 2013) to reduce recruitment expenses, the high fixed costs of training new workers, and the high cost of labor mobility (Bryan, 2007). A similar approach is also induced by adverse selection models of the labor market, where many firms attract hard-working and productive workers by offering extra hours (Sousa-Poza & Ziegler, 2003). In contrast, constraints related to job changes and limited work hours restrict individuals from participating in the labor market (Stewart & Swaffield, 1997). Hence, both of these constraints further compel workers to stay with limited offers and hard jobs (Bryan, 2007; Stewart & Swaffield, 1997), which also reduces workers’ ability to utilize the intensive margin of labor supply (Bui & Ume, 2020). In the labor market, the inequality of productivity across work hours causes both workers and employers to focus more on which tasks will be assigned in the first few hours of an employee’s workday (Barzel, 1973).1 Firms usually set up tasks and schedules at the start of the day or week, which obviates the laborer’s objective of working more hours. The worker’s aim to provide extra hours is also unprofitable for firms because workers’ productivity decreases with increasing fatigue (Bryan, 2007). Although the monopsony buying power of firms means that workers can work more hours, they are paid a lower hourly wage rate, even with their increasing effort (Naylor, 2002). In addition to the hour constraints and worker productivity, fewer working hours among credit-constrained individuals also result from the contract model, where wages and employment are determined by long-term agreements for insurance reasons. However, wages in such a system do not equal marginal productivity or efficient hours (Bryan, 2007). Hence, the literature provides significant evidence for the impact of credit constraint on the extensive and intensive margin of labor supply decisions.

Given the importance of the credit market, this study analyzes the impact of credit constraints on labor supply decisions and contributes to the literature in the following ways. First, we employ a direct and observed measure of the credit market, i.e., credit constraint, and analyze its impact on labor market outcomes. Previous studies have used traditional measures of liquidity constraints, such as initial wealth and inheritance wealth, or focused mainly on the subjective measures of financial imperfections or developed proxies, such as home equity and lottery income. For example, Bernstein (2021) used negative home equity, while Cesarini et al. (2017) used monetary prizes from Swedish lottery players as a measure of wealth to analyze its impact on labor supply. More recently, Kumar and Liang (2024) analyzed the US labor market and considered the 1998 and 2003 constitutional amendments in Texas for home equity borrowing as a measure of credit access. Lastly, Benito and Saleheen (2013) considered the subjective behavior of financial shocks and could not analyze the underlying sources and reasons for financial shocks. We overcome this problem by exploring the observed credit constraint.

Second, we do not focus solely on the rejected borrowers as a measure of credit constraint but also incorporate other measures of credit constraint, such as applying for credit but not giving much (underfunded borrowers) and being discouraged because of high interest rates in the market. We use these measures and analyze the impact on both labor force participation and also the working hours supplied, referred to as the extensive and intensive margins of the labor market, respectively. Thus, our measurement is based on the demand or borrower side of the credit market and contributes to the existing literature, which has primarily focused on the supply side of the credit market. For example, Bui and Ume (2020) use the bank branching deregulation as a proxy for credit constraint across states, which is based on the supply side of credit channels. Third, we analyze the impact of credit constraints and their various measures on job search outcomes, which is a measure of the time taken to find reemployment. We explore whether credit-constrained individuals are more likely to search for jobs compared with those who are not constrained and accept reemployment. Fourth, we also extend the credit constraint and labor supply decisions by considering wages and choice of occupation among households.2

Hence, this study connects the dots between the financial and labor markets by analyzing the impact of financial shocks arising from credit constraints on the labor market and household decisions to supply labor. Both credit constraints and labor market decisions can be correlated with the unobserved individual characteristics, and reverse causality may be present. To address endogeneity, we use the distance from the borrower to the lender’s location as an instrumental variable for credit constraints. The choice to work more and the decision to participate in the labor force are usually interdependent and depend on the nature of occupational risk, such as the individual choice to join risky self-employment activities or paid employment opportunities.3 The empirical findings confirm our hypothesis that the credit constraint increases the probability of households supplying more labor. Other measures of credit constraint also significantly induce similar findings.

The rest of the study is organized as follows: Section 2 discusses the literature on financial shocks and labor market decisions. Section 3 provides the data used in the study along with a description of the variables. Section 4 describes an estimation technique. Section 5 summarizes the results. Finally, Section 6 concludes the study by analyzing the impact of credit constraints on households’ decisions to work.

2. Financial Shocks and Labor Market

2.1. Liquidity Constraint, Job Search Time, and Labor Supply

The impact of financial shocks on household behavior and its influence on labor decisions has become a key focus in the labor and finance literature. The seminal work of Deaton (1991) and its extension by Attanasio and Weber (2010) argue that liquidity constraints restrict consumers’ ability to borrow from the market, undermining the permanent income and consumption smoothing model. Bernhardt and Backus (1990) introduce the borrowing constraint in a theoretical model of the labor market and show that the choice of occupation is influenced by the wealth status of the individual. Cheng and French (2000) extend the literature by evaluating the borrowing constraint and uncertainty about wages. They concluded that households are unable to borrow against future labor income. Evans and Jovanovic (1989) modify the labor market and the financial constraint debate by considering liquidity constraint and self-employment together. The literature has also analyzed that microcredit opportunities, or small-scale loans, also benefit people in smoothing consumption. Friebel and Giannetti (2009) consider the role of credit limits in the consumption smoothing process and conclude that people with higher credit limits prefer to join less secure jobs. Pitt and Shahidur (1996) analyze the case of Bangladesh and examine how microcredit opportunities benefit households in several ways.

On the other hand, the decision to work or supply additional hours in the labor market has been widely discussed in the public economics literature, where tax changes affect labor decisions (Brewer et al., 2010; Laroque, 2005). Only a few studies have considered the link between liquidity constraints and labor supply. The work of Breunig and Cobb-Clark (2005) and Kiely et al. (2015) analyzes the moderate and severe economic/psychological impacts of liquidity constraints on the economic and financial components of households. Further, Rossi and Trucchi (2016) contribute by using the instrumental variable method and household data from the Italian survey and conclude that liquidity constraints have a negative relationship with labor supply. Another set of studies, including the work of Bottazzi (2004), Cao (2017), and Lusardi and Mitchell (2017), extends the literature on mortgage-related liquidity constraints and their impact on labor supply. They concluded that mortgage-related constraints have positive impacts. Boca and Lusardi (2003) and Kumar and Liang (2024) find similar results regarding the negative effects of increased credit availability on labor supply.

Basten and Telle (2012); Chetty (2008); Corsini (2012); Gerards and Welters (2020) analyze the importance of credit constraints and recognize them as a component of asymmetric information. Additionally, the literature conclusively argues that financial constraints in the credit market act as a buffer in the labor market. However, the literature has completely overlooked the analysis of the impact of credit constraints on the labor market outcome. Hence, we develop a research hypothesis and argue that credit constraints result in higher labor force participation. Based on the literature, we propose the following hypotheses:

Hypothesis 1.

Credit constraints are positively associated with labor force participation.

We consider that less availability of credit or being turned down for credit applications discourages individual consumption in the current period when they are unemployed. This reduces their ability to use cash to meet their needs. Hence, they experience “hardship” and “cash flow problems” (Kiely et al., 2015). As a result, the constraint of credit and hardship encourages people to search for jobs (Gerards & Welters, 2020) and join the labor force to smooth their consumption.

Secondly, we extend our discussion on labor force participation and consider the effectiveness of the job search process when individuals are unemployed and experiencing financial hardship. Gerards and Welters (2020) point out that unemployed individuals search for jobs and are likely to accept employment when faced with liquidity constraints. To explore this analysis, we introduce the short-term outcome of the job search, or time to reemployment, as a proxy for job search effectiveness due to credit constraints. Gerards and Welters (2020) concluded that individuals who are liquidity-constrained are reemployed later than those who are without such constraint.

Hypothesis 2.

The positive association between credit constraints and short-run job search outcomes (defined by search time to reemployment) is evident because credit-constrained individuals are more likely to search for jobs and accept reemployment.

2.2. Wealth, Finance, and Work Hours

In the labor market, individual financial constraints, hardship, and cash flow problems are not the only factors determining work hours. Firm production capacity, hiring costs, and financial capacity also determine work hours for employees. Benito and Saleheen (2013) argue that firms usually pay more fixed costs in terms of training and social insurance, which can reduce their profitability, prompting them to offer more working hours. Hence, individuals can expect different work hours from their employer than they would. On the labor supply side, Bryan (2007) argues that high mobility costs force workers to provide more labor hours. In the presence of credit constraints, the workers always seek financial compensation and have no option except to provide more labor to smooth their consumption. Rossi and Trucchi (2016) are known for pioneering work that introduces the liquidity constraint in the labor supply model. They used data from the Italian Survey of Household Income and Wealth (SHIW) and concluded that liquidity constraints induce workers to supply four more hours of work. Bottazzi et al. (2007) also develop the link between the intensity of labor supply and mortgage debt. Literature also highlights the importance of family wealth in determining the individual decision to supply labor. Zhao and Burge (2017) analyze the role of family wealth and conclude that higher family wealth periods provide less labor force participation. In extending the literature, Bui and Ume (2020) introduce bank branch deregulation as a solution to liquidity constraints and conclude that worker hours decrease by 0.5. In this context, our hypothesis can be as follows:

Hypothesis 3.

There is a positive relationship between credit constraints and working hours.

Liquidity constraints can shape household decisions about labor supply (Rossi & Trucchi, 2016). Individuals whose credit applications have been turned down are more likely to work more to smooth their consumption or savings trajectories. Hence, credit-constrained individuals prefer to extend their labor supply as a way to reduce the negative effects of credit constraints.

3. Data and Variables

The empirical analysis uses the cross-sectional data from the Survey of Consumer Finance (SCF). This is a triennial cross-sectional survey of US families, and its 2022 wave contains the information of 4,595 families with a sample size of 22,975 individuals. There are a few advantages to using the SCF;4 First, the survey has complete information on the family balance sheet, finances, income, demographic characteristics, labor status, and other aspects at the individual level. Secondly, it gives direct and detailed descriptions of credit access for US households. It also has information about credit constraints, such as “being turned down for credit applications” and “not given as much credit as applied for”. It further provides sufficient information about the discouraged borrowers. Thirdly, SCF has an advantage over the Survey of Household and Economic Decisionmaking (SHED) and other finance surveys in terms of complete information on the labor market, labor participation, and decisions to work. It also provides the labor market data about the exact working hours and time to reemployment for unemployed individuals, which is our interest to explore.

In

Table 1. Description of Variables.

Variables	Definitions	Mean
Labor Supply Decision
Extensive Margin
Labor force participation	=1 if individual labor force status is employed	75.98
Time to search employment (c)	Number of weeks it took the unemployed to find employment	19.04
Intensive Margin (c)	Number of hours worked in last week	46.55
Credit Request:
Credit Constrained	=1 if credit application was constrained in any way	17.74
Rejected	=1 if credit application has been turned down	8.82
Underfunded	=1 if request for credit was approved but underfunded	2.47
Discouraged	=1 if think not to be approved and not applied	3.88
Price out	=1 if observe high interest rate for loan and not applied	2.56
Rejected (revised) Borrower	=1 if was still rejected after multiple attempts	4.01
Income Sources
wages	Average wage per hours	42.77
Log (savings)	The logarithmic value of savings in savings or money market acount	5.07
Risk:
Occupational Risk	=1 if household runs a business or work as self-employed	2.73
Wealth:
Auto Ownership	=1 if household owns their own automobile	85.27
Home Mortgage	=1 if household has mortgage on their home	2.71
Socioeconomic Variables:
Age (c)	The avarge agae of the individual	51.71
Household Size	The avarage family size	2.36
Gender (omitted: Male)
Female	=1 if head of household is female	23.90
Martial Status (omitted: Single)
Married	=1 if head of household is married	54.60
Separated	=1 if head of household is separated	17.13
Widowed	=1 if head of household is widowed	6.72
Race and Ethnicity (omitted: White)
Black	=if head of household is Black	15.60
Asian/Indian/Hawaiian	=if head of household is Asian/American Indian/Native Hawaiian	8.78
Hispanic/Latino	=if head of household is Hispanic	13.60
Education (omitted: <High School)
High School	=if head of household has high school diploma only	19.68
Bachelors Degree	=if head of household has a bachelors degree	47.21
Post Graduate	=if head of household has some post-graduate education	23.76

Notes: c stands for continuous variable.

, the credit constraint is measured as households that (i) have applied for bank credit but “their credit applications are turned down” (rejected) or “could not get as much credit as requested” (underfunded). There are 8.82% and 2.47% of households contributing to both rejected and underfunded categories, respectively. The credit constraint also includes those households (ii) who could not apply because “the interest rate was too high” and/or “did not think of getting approved”. We call them discouraged and price out borrowers with percentage values of 3.88 and 2.56, respectively. Households that observe at least one type of financial constraint are considered credit-constrained and account for 17.74% of our sample. Lastly, we also introduce another variation of rejected borrowers and call it rejected–revised borrowers. This revised definition of rejected borrowers includes those individuals who were initially rejected or underfunded but reapplied and were rejected again. 4.01% of households fall into this category. Jappelli (1990) introduced this variable in his credit constraint work and called it discouraged borrowers. His work used the 1982 SCF wave and found that 12.4% (370 out of 2971) fall into this category. We noticed that the percentage of these borrowers has decreased in the last 40 years.

Our variables of interest include the intensive margin and the extensive margin of labor supply, which are used as dependent variables in two regression equations. The proxy for intensive margin is the hours of work5 in a normal week by individual i in a year t. Table 1 shows that 46.55 is the average work hour per week for people 25 years or older. The proxy for extensive margin is labor force participation, measured by the labor force status of an individual i in a year t with a percentage value of 75.98. Many studies, including Blundell et al. (2011) and Bui and Ume (2020), have used both variables as a proxy for intensive and extensive margins. We also introduce another proxy for an extensive margin and replace our dependent variable with a short-run job search outcome or the time to find reemployment. This is measured by the number of weeks it took the unemployed to find reemployment. In our sample, it takes, on average, 19.04 weeks for unemployed individuals to find reemployment and start work. Gerards and Welters (2020) used this in their work and found that the average time is more than 20 weeks among Australian workers using HILDA survey data. This indicates that US (unemployed) workers can find their reemployment earlier than Australian workers.

We also explore the impact of wages on labor supply and evaluate the labor–leisure decisions of individual workers, which can help us conclude the substitution and income effects on labor supply. In this context, we consider the hourly wage of individuals and use it in our credit constraint model. Our sample indicates that the average wage of the respondents is $42.77 per hour (with a logarithmic value of 2.07), very similar to the work of Rossi and Trucchi (2016) in the case of Italy. We also use its squared form to analyze the non-linear relationship between wages and both labor force participation and labor supply. Additionally, non-labor income or savings plays a crucial role in the labor supply model. Block and Heineke (1973) and Chiu and Eeckhoudt (2010) argue that people usually increase their labor supply in response to higher non-labor income or savings because they use it as a hedge against uncertainty. We also expect savings to play an important role in individual decisions to join the labor force or supply more labor in the presence of credit constraints. Hence, we introduce non-labor income or savings for individual i at time t. This is considered as individual savings in savings or money market accounts that have a logarithmic value of 5.07 in our sample.

Additionally, the labor decision to supply more or less labor or enter the labor market, either directly or indirectly, depends on the choice and nature of occupations. In defining occupational income risk and its impact on labor supply decisions, we follow the work of Giannetti (2011) and use self-employment as a proxy. The literature argues that jobs based on self-employment and entrepreneurship are known to be riskier compared with paid employment opportunities such as jobs in private or public sectors. Hence, paid employment is considered less risky because employment protection measures often facilitate and provide security to paid workers. Against this background, this study uses occupational income risk, taking the value of 1 if individual i in year t is self-employed and 0 otherwise. In addition, individual wealth is measured by home ownership or auto ownership, which are important determinants of the individual labor supply response. We incorporate both variables into our credit constraint model. Our sample indicates that there are 2.71% and 85.27% homeowners and auto owners, respectively. Lastly, we also control for age, family size, educational dummies, sex, race/ethnicity, and marital status.

4. Econometric Methodology

This paper estimates the impact of the credit constraint and its other measures on labor force participation, time to find reemployment, and the intensity of work hours. The underlying equations are given as follows:

$$LF{S}_i=X_i^{\prime }\theta +\beta C{C}_i+\gamma O{R}_i+\zeta Wag{e}_i+u_i$$

(1)

$$JS{O}_i=X_i^{\prime }\theta +\beta C{C}_i+\gamma O{R}_i+\zeta Wag{e}_i+v_i$$

(2)

$$W{H}_i=X_i^{\prime }\theta +\beta C{C}_i+\gamma O{R}_i+\zeta Wag{e}_i+e_i$$

(3)

where $LF{S}_i$ is the employment status of an individual i and used as a dependent variable in Equation (1). It is a binary dummy variable and takes the value equal to 1 if the individual labor force status is known as employed and 0 otherwise. $JS{O}_i$ is the short-term job search outcome of individual i in Equation (2). This considers the time to find employment for an unemployed individual and takes values from 1 to 52 (in weeks). While $W{H}_i$ is the working hour supplied by individual i in Equation (3). This is a continuous variable with a median value of $13.89. $X_i^{\prime }$ is the matrix of covariates and includes the logarithm of savings, wealth variables, age of individuals, age-squared, categories of education, race and ethnicity, dummy variables for gender, and marital status. $C{C}_i$ is the dummy variable and is equal to 1 if the individual is credit-constrained. Credit constraint is the combination of four categories and includes those individuals whose (i) credit applications have been turned down (rejected) or (ii) not given as much credit as applied for (underfunded) and did not apply because (iii) they think they would not be approved (discouraged) or (iv) their interest rate is too high (priced out). $O{R}_i$ is also a dummy variable and equal to 1 if an individual i is engaged in a risky occupation such as self-employment or entrepreneurship. $Wag{e}_i$ is a continuous variable in our credit constraint model and captures the substitution or income effect of labor supply, where $\beta$, $\gamma$, and $\zeta$ are the coefficients of credit constraint ($C{C}_i$), occupational income risk ($O{R}_i$), and per-hour wage ($Wag{e}_i$). $u_i$, $v_i$, and $e_i$ are the error terms.

The estimated $\beta$ from the OLS can be biased (Rossi & Trucchi, 2016) because unobserved characteristics, such as an individual’s attitude toward leisure, less flexibility in labor–leisure trade-off, time preference, the worker’s own characteristics such as concerns for family time, permanent disability, and individual status in working hours, can determine the labor supply, which further increases the likelihood of being credit constrained. Secondly, endogeneity can also arise when credit constraints are not exogenous and are correlated and determined by other factors such as the income of the individuals, living status, and personal characteristics. In addition, banks always screen potential applicants by evaluating their personal characteristics, which can result in credit constraints. Third, reverse causality can exist when an individual works less or contributes fewer hours of work, which encourages the lender to offer fewer loans or turn down the credit application. Hence, all of the above factors push the effect of $\beta$ downward, resulting in biased estimators, and the credit constraint variables may be measured with error. The same is true for the Probit model.

To address the endogeneity, an instrumental variable is introduced in all three empirical models. The rationale behind the instrumental variable is clear: it provides an alternative source and empowers individuals to fully access the credit market to smooth consumption. A shorter distance to financial institutions may reduce the credit constraint, and we expect the probability of being credit-constrained to decrease with the development of financial institutions closer to the borrower’s location. It is also expected that households living farther from financial institutions have less access to credit. Hence, we utilize the distance of the financial institution from the borrower’s location as an instrument. This has been used in the literature (Cai et al., 2018). Furthermore, exclusion restrictions are satisfied, and the instrument, “distance to lender location”, does not correlate with the dependent variable other than the credit constraint. From this perspective, we create a dummy variable6 taking the value of one if the distance7 between the individual or borrower’s location and the financial institutions is less than half a mile and 0 otherwise. SCF considers seven broader categories of financial institutions that include (i) a commercial bank (8.71%), (ii) a savings and loan or savings bank (4.50%), (iii) a credit union (1.86%), (iv) a mortgage company (1.11%), (v) a finance or loan company (0.67%), (vi) a brokerage (0.37%), and (vii) something else (0.31%). This shows that 8.71% of the individuals are less than half a mile from commercial banks, and 4.50% of the individuals are less than half a mile from savings and loan or savings banks. Brokerage and other categories are farther from borrower locations, making them the least accessible institutions, with only 0.37% and 0.31% of individuals, respectively, having access.

5. Results and Discussions

We start working with the raw data and perform a two-sided t-test on credit constraints and labor market variables. We analyze how credit constraints differ across labor force participation, working hours, labor and non-labor income, and occupational risk. Our first variable of interest is the extensive margin, as measured by labor force participation.

Table 2. (Differences in) means by credit constraint for extensive margin, search intensity, intensive margin, labor and non-labor income, and occupational risk.

Two-Sided T-Test	CreditC	RejB	UnderFB	DisB	PriceOB	RejRB
Extensive Margin
(i) Labor Force Participation	0.55 ***	0.65 ***	0.73 ***	0.71 ***	0.73 ***	0.71 ***
Short-run Job Search Outcome
(ii) Time to Employment	1.90 ***	1.99 ***	2.05 ***	2.04 ***	2.05 ***	2.04 ***
Intensive Margin
(i) Number of Hours Worked	46.37 ***	46.45 ***	46.52 ***	46.52 ***	46.52 ***	46.51 ***
Labor and non-labor income
(i) Wages (log)	1.89 ***	1.98 ***	2.04 ***	2.03 ***	2.04 ***	2.03 ***
(ii) Savings (log)	4.89 ***	4.98 ***	5.04 ***	5.03 ***	5.04 ***	5.03 ***
Occupational Risk	−0.15 ***	−0.06 ***	0.001 *	−0.01 ***	0.001	−0.12 ***

Notes: (i) *** p < 0.01, * p < 0.1. (ii) CreditC: credit constraint, RejB: rejected borrower, UnderFB: underfunded borrower, DisB: discouraged borrower, PriceOB: priced-out borrower, RejRB: rejected revised borrower.

shows that labor force participation among credit-constrained individuals significantly increases compared with non-credit-constrained individuals. The impact becomes stronger when using an alternate form of credit market imperfection.

Our second variable of interest is job search outcome, which is defined as the number of weeks it takes an unemployed individual to find employment. The employment period ranges from 0 to 52, where 0 indicates that they found employment in the first week, and the latter indicates that it can take up to 52 weeks to have employment. The findings show that the time to find reemployment among credit-constrained individuals is significantly longer than for non-constrained individuals, specifically those experiencing other severe forms such as underfunded and priced-out borrowers. Additionally, our findings show that the intensive margin, as measured by the number of hours worked in the last week, is positive and significant for credit-constrained individuals. Its intensity and significance remain robust across different forms of credit constraint. Table 2 also shows that wages and savings are significant and positive for credit-constrained individuals and their different forms. Occupational risk is significant but has not been consistent in sign when analyzed with other forms of credit constraint.

5.1. Analysis of Job Search Outcome

In

Table 3. Labor force participation—probit and IV probit regression models.

	(1)	(2)	(3)	(4)	(5)	(6)
	Probit	Probit	First Stage	Second Stage	First Stage	Second Stage
Credit Constrained	−0.014 **				0.743 ***
	(0.006)				(0.287)
Rejected Borrower		−0.007				1.568 *
		(0.008)				(0.810)
Distance to Lender			0.041 ***		0.019 **
			(0.010)		(0.008)
Wage	0.150 ***	0.150 ***	−0.009 *	0.156 ***	0.015 ***	0.126 ***
	(0.004)	(0.004)	(0.005)	(0.006)	(0.004)	(0.014)
Wage Sq.	−0.014 ***	−0.014 ***	−0.001	−0.013 ***	−0.004 ***	−0.008 **
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.003)
Savings	0.001 **	0.001 **	−0.010 ***	0.009 ***	−0.006 ***	0.011 **
	(0.001)	(0.001)	(0.001)	(0.003)	(0.001)	(0.005)
Occupational Risk	0.224 ***	0.225 ***	−0.033 *	0.250 ***	−0.008	0.238 ***
	(0.021)	(0.021)	(0.017)	(0.027)	(0.014)	(0.031)
Age	0.020 ***	0.020 ***	0.005 **	0.016 ***	0.003 *	0.014 ***
	(0.002)	(0.002)	(0.002)	(0.003)	(0.002)	(0.005)
Age Sq.	−0.000 ***	−0.000 ***	−0.000 ***	−0.000 ***	−0.000 ***	−0.000 ***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
Auto Ownership	0.047 ***	0.047 ***	−0.034 ***	0.075 ***	0.012	0.032 **
	(0.007)	(0.007)	(0.009)	(0.015)	(0.007)	(0.015)
Home Mortgage	0.020	0.019	0.008	0.013	−0.001	0.020
	(0.016)	(0.016)	(0.020)	(0.022)	(0.016)	(0.030)
Family Size	−0.003	−0.003	0.017 ***	−0.016 ***	0.007 ***	−0.014 **
	(0.002)	(0.002)	(0.003)	(0.006)	(0.002)	(0.007)
Female	−0.046 ***	−0.047 ***	0.030 ***	−0.068 ***	0.001	−0.048 ***
	(0.007)	(0.007)	(0.009)	(0.013)	(0.007)	(0.013)
Married	0.028 ***	0.029 ***	−0.048 ***	0.065 ***	0.004	0.023
	(0.008)	(0.008)	(0.010)	(0.018)	(0.008)	(0.015)
Separated	0.012	0.011	0.053 ***	−0.027	0.049 ***	−0.066
	(0.008)	(0.008)	(0.010)	(0.018)	(0.008)	(0.042)
Widow	−0.041 ***	−0.042 ***	0.030	−0.063 ***	0.069 ***	−0.149 **
	(0.014)	(0.014)	(0.019)	(0.022)	(0.015)	(0.062)
High School	0.036 ***	0.037 ***	−0.022 *	0.052 ***	0.036 ***	−0.020
	(0.009)	(0.009)	(0.012)	(0.014)	(0.010)	(0.034)
Bachelor Degree	0.030 ***	0.032 ***	−0.085 ***	0.093 ***	0.005	0.022
	(0.009)	(0.009)	(0.012)	(0.027)	(0.009)	(0.018)
Post Graduate	0.093 ***	0.094 ***	−0.104 ***	0.170 ***	−0.003	0.097 ***
	(0.011)	(0.011)	(0.013)	(0.033)	(0.010)	(0.019)
Black	−0.018 **	−0.019 ***	0.163 ***	−0.140 ***	0.083 ***	−0.150 **
	(0.007)	(0.007)	(0.009)	(0.048)	(0.007)	(0.069)
Hispanic/Latino	0.010	0.010	0.034 ***	−0.018	0.011	−0.009
	(0.008)	(0.008)	(0.009)	(0.015)	(0.007)	(0.017)
Asian/Indian/Hawaiian	−0.030 ***	−0.030 ***	0.006	−0.038 ***	0.014 *	−0.055 ***
	(0.009)	(0.009)	(0.010)	(0.012)	(0.008)	(0.020)
Constant			0.269 ***		0.036
			(0.054)		(0.043)
Observations	16,386	16,386	16,386	16,386	16,386	16,386
Wald Chi-squared				12,967.09 (0.00)		19,355.40 (0.00)
Pseudo R-squared	0.426	0.425
Endogeneity test
Wald test of exogeneity				15.07 (0.00)		12.41 (0.00)

Notes: (i) Standard errors in parentheses. (ii) *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

, we report the regression results of the probit model for the impact of credit constraint on labor force participation, a measure of extensive margin. In columns 1 and 2, the impact of credit constraint and its other measures on rejected borrowers is negative on labor force participation. The probit model provides biased results, and our main challenge is to identify the causal effect of the credit constraint on labor force participation. We attempt to alleviate the endogeneity problem by finding an instrumental variable. In our analysis, we consider the distance to the financial institution from the borrower’s location as an instrument and regress the IV-probit model. In columns 3 and 5, the first-stage results are reported. The instrumental variable is statistically significant. The results of the second stage are reported in columns 4 and 6, which show that households subject to credit constraints are significantly more likely to increase labor force participation. In column 6, we checked robustness by replacing the credit constraint variable with the rejected borrowers (turned-down credit applications). The relationship between rejected borrowers and participation in the labor force is positive and statistically significant.

Baseline findings show that suffering from credit constraints significantly increases the probability of extensive margin of the labor market by about 74.3 percentage points.8 These findings are consistent with the work of Kumar and Liang (2018) and Bui and Ume (2020). They used a similar framework and concluded that credit expansion leads to a reduction in working hours when bank branch deregulation is introduced. The probability value of extensive margin increases even further when we replace the credit constraint with the rejected borrowers. This means that the inability of individuals to borrow from the credit market encourages them to participate more or start working if they are unemployed. This also confirms that credit-constrained individuals begin to experience more hardship and cash flow problems, which discourage them from enjoying the benefits of being unemployed. In columns 4 and 6, we also consider the impact of wages, wage-squared, and savings on the individual’s choice to participate in the labor market. Our findings indicate that people start participating more significantly in the labor market due to higher wages and savings. This suggests that higher wages attract people to give up their leisure for labor participation. In addition, increases in wages exhibit an inverse U-shaped relationship with labor force participation. Individual savings also induce a positive relationship with participation in the labor force. Occupational risk, measured by self-employment, significantly increases participation in the labor force in both models. This suggests that self-created businesses and self-employment encourage individuals to maximize their return by increasing their participation in the labor market. Additionally, we report the results of the endogeneity test at the bottom of columns 4 and 6. The test results reject the null hypothesis that both credit constraints and rejected borrowers are exogenous. This confirms that the findings of the probit model differ significantly from those of the IV probit model in terms of sign and significance. Our findings show that credit market imperfections play a crucial role in increasing labor force participation.

To validate the findings, we performed the robustness analysis employing alternative definitions of credit constraint. In

Table 4. Alternate definitions: IV probit regression models of labor force participation.

Variables	(1)	(2)	(3)	(4)
Constraint 1	2.598 *
	(1.352)
Constraint 2		4.703
		(4.166)
Constraint 3			2.480 *
			(1.298)
Constraint 4				2.482 *
				(1.462)
Wage	0.145 ***	0.258 ***	0.156 ***	0.157 ***
	(0.007)	(0.097)	(0.007)	(0.009)
Wage Sq.	−0.011 ***	−0.029 **	−0.014 ***	−0.014 ***
	(0.002)	(0.014)	(0.001)	(0.002)
Savings	0.002 **	0.016	0.003 **	0.007 **
	(0.001)	(0.013)	(0.001)	(0.003)
Occupational Risk	0.280 ***	0.217 ***	0.235***	0.272 ***
	(0.041)	(0.049)	(0.029)	(0.043)
Observations	16,386	16,386	16,386	16,386
Wald Chi-squared	17,206 (0.00)	27,529.6 (0.00)	17,868.7 (0.00)	21,540.7 (0.00)
Wald test of exogeneity	10.63 (0.00)	7.47 (0.00)	10.83 (0.00)	10.36 (0.00)

Notes: (i) Also included age, age squared, auto ownership, home mortgage, family size, gender, categories of education and race. Standard errors in parentheses. (ii) *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level. (iii) 1-Stage F-Statistics results are significant and 1-Stage results are available. (iv) Constraint 1: applied for credit but not given much credit (underfunded). (v) Constraint 2: did not apply for credit because they thought it would not be approved (discouraged). (vi) Constraint 3: did not apply for credit because interest rate too high (priced out). (vii) Constraint 4: rejected–revised borrowers.

, we exclude the variable “credit constraint” and construct additional measures of credit constraint, including the following: Constraint 1: applied for credit but received insufficient amounts (underfunded). Constraint 2: did not apply because they thought approval was unlikely (discouraged). Constraint 3: interest rate too high; priced out. Constraint 4: individuals still credit-constrained after reapplying; rejected revised borrowers. The empirical findings show that various forms of credit constraint significantly increase the probability of participation in the labor force. Constraint 2 does not provide a significant impact along with its value, likely due to the low number of respondents in this category. Among the constraints, the magnitude of the coefficients is smallest for Constraint 3 (priced-out borrowers) but significantly higher for Constraint 1 (underfunded). One possible explanation is that individuals may lack sufficient information about interest rates due to a lack of financial literacy. However, those who have adequate information may receive partial funding when applying for a loan. Furthermore, the findings related to both wages and savings significantly contribute to increased participation in all specifications of credit constraints. All other variables are robust and show similar signs and significance.

5.2. Analysis of Job Search Outcome

The literature points out that credit constraints increase hardship and cash flow problems among unemployed individuals, encourage them to seek employment, and increase participation in the labor force. In this context, reemployment occurs through two possible channels that include (i) an increase in the job search intensity among constrained individuals, measured by the number of hours spent on job search, and (ii) the job search outcome, meaning that credit-constrained individuals require more time to find reemployment. The SCF does not collect data on hours spent searching for a job, but it has data on the time to find employment. Therefore, we replace our proxy for the extensive margin and introduce short-term job search outcomes as a dependent variable, measured by the time taken to secure reemployment. In

Table 5. Short-run job search outcome (time to find reemployment)-2SLS models.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
	First Stage	Second Stage	First Stage	Second Stage	First Stage	Second Stage
Credit Constrained		2.012 **
		(0.940)
Rejected Borrower				27.46
				(94.32)
Constraint 1						2.569 **
						(1.055)
Distance to Lender	0.0536 ***	0.00393	0.0420 ***
	(0.0174)	(0.0135)	(0.00758)
Wage	−0.00519	−0.119 ***	0.0131 ***	−0.489	0.00104	−0.132 ***
	(0.00390)	(0.0124)	(0.00304)	(1.236)	(0.00170)	(0.00998)
Wage Sq.	−0.00148 *	0.0152 ***	−0.00317 ***	0.0993	−0.000716 **	0.0140 ***
	(0.000798)	(0.00266)	(0.000621)	(0.299)	(0.000348)	(0.00215)
Savings	−0.00930 ***	0.0101	−0.00513 ***	0.132	−0.000246	−0.00802 ***
	(0.000505)	(0.00885)	(0.000393)	(0.484)	(0.000220)	(0.00131)
Occupational Risk	−0.0232	0.0993 **	0.00242	−0.0136	−0.0197 ***	0.103 **
	(0.0144)	(0.0465)	(0.0112)	(0.389)	(0.00630)	(0.0416)
Age	0.000103	−0.00546 **	−0.00141 **	0.0336	−0.000846 **	−0.00308
	(0.000907)	(0.00263)	(0.000705)	(0.134)	(0.000395)	(0.00246)
Age Sq.	−3.39 × ${10}^{-5}$ ***	1.31 × ${10}^{-5}$	−8.12 × ${10}^{-6}$	0.000168	4.15 × ${10}^{-6}$	−6.58 × ${10}^{-5}$ ***
	(8.78 × ${10}^{-6}$)	(4.10 × ${10}^{-5}$)	(6.83 × ${10}^{-6}$)	(0.000791)	(3.83 × ${10}^{-6}$)	(2.27 × ${10}^{-5}$)
Auto Ownership	−0.0293 ***	−0.0274	0.000219	−0.0925	0.0161 ***	−0.128 ***
	(0.00704)	(0.0346)	(0.00548)	(0.151)	(0.00307)	(0.0245)
Home Mortgage	0.00232	−0.0341	0.00664	−0.212	−0.00645	−0.0128
	(0.0146)	(0.0424)	(0.0113)	(0.707)	(0.00636)	(0.0376)
Family Size	0.0170 ***	−0.0215	0.00499 ***	−0.124	0.00462 ***	0.000877
	(0.00222)	(0.0172)	(0.00173)	(0.473)	(0.000967)	(0.00747)
Female	0.0187 ***	0.0364	−0.00675	0.259	0.00631 **	0.0578 ***
	(0.00703)	(0.0264)	(0.00547)	(0.659)	(0.00306)	(0.0189)
Married	−0.0523***	−0.0470	0.00184	−0.203	−0.00446	−0.141 ***
	(0.00829)	(0.0550)	(0.00645)	(0.246)	(0.00362)	(0.0217)
Separated	0.0413 ***	−0.206 ***	0.0440 ***	−1.330	0.00661 *	−0.140 ***
	(0.00828)	(0.0455)	(0.00644)	(4.149)	(0.00361)	(0.0222)
Widow	−0.00925	−0.212 ***	0.0361 ***	−1.221	0.00795	−0.251 ***
	(0.0119)	(0.0356)	(0.00925)	(3.408)	(0.00519)	(0.0314)
High School	−0.0663 ***	0.0714	0.0283 ***	−0.841	0.00540	−0.0759 ***
	(0.00966)	(0.0682)	(0.00752)	(2.683)	(0.00421)	(0.0253)
Bachelor Degree	−0.112 ***	0.154	0.00425	−0.189	0.00558	−0.0863 ***
	(0.00919)	(0.108)	(0.00715)	(0.450)	(0.00401)	(0.0242)
Post Graduate	−0.129 ***	0.101	−0.00752	0.0481	0.0125 ***	−0.191 ***
	(0.0102)	(0.123)	(0.00794)	(0.734)	(0.00445)	(0.0296)
Black	0.159 ***	−0.168	0.0793 ***	−2.026	0.0288***	0.0787 **
	(0.00725)	(0.151)	(0.00564)	(7.484)	(0.00316)	(0.0356)
Hispanic/Latino	0.0352 ***	0.0327	0.00812	−0.119	0.00684 **	0.0859 ***
	(0.00785)	(0.0404)	(0.00611)	(0.786)	(0.00342)	(0.0214)
Asian/Indian/Hawaiian	0.000899	0.105 ***	0.00700	−0.0857	−0.00722 *	0.125 ***
	(0.00856)	(0.0248)	(0.00666)	(0.690)	(0.00373)	(0.0229)
Constant	0.410 ***	0.287	0.167 ***	−3.471	0.0240 **	1.051 ***
	(0.0242)	(0.391)	(0.0189)	(15.75)	(0.0106)	(0.0666)
Observations	22,975	22,975	22,975	22,975	22,975	22,975
Wald Chi-squared		1641.73 (0.00)		30.34 (0.00)		2122.65 (0.00)
R-squared	0.143		0.058		0.014
1st Stage F-Statistics	191.30 (0.00)		70.84 (0.00)		15.93 (0.00)
Endogeneity Test
Durbin Chi-Sq.		5.52 (0.01)		7.08 (0.00)		6.40 (0.01)
Wu-Hausman		5.52 (0.01)		7.08 (0.00)		6.40 (0.01)

Notes: (i) Standard errors in parentheses. (ii) *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level. (iii) Constraint 1: Applied for credit but not given much credit (underfunded).

, we test our hypothesis (number 2) and check it with various forms of credit market variables that include (i) credit-constrained, (ii) rejected borrowers, and (iii) underfunded borrowers.

Our findings conclude that credit-constrained individuals take significantly more time to find reemployment compared with unconstrained individuals. Several factors contribute to delayed reemployment. First, an individual’s socio-economic conditions can hinder their ability to find new opportunities, with marginalized groups facing greater challenges. Second, temporary disabilities may delay the job search process, extending the time needed to secure reemployment. Lastly, a lack of education, skills, or training can severely limit a person’s ability to compete in the labor market. These factors negatively impact individuals’ ability to secure employment and contribute to longer periods of unemployment.

In extending our analysis, we check the robustness of our results by replacing the credit constraint with rejected borrowers and find a positive but insignificant relationship. However, the findings become significant in model 3 (column 6) when underfunded individuals are analyzed, who apply for credit but cannot obtain much. Broadly, we can conclude that credit-constrained individuals take a few more weeks in the job search process and are employed later than those who are not credit-constrained. The possible reason is that their unemployment period makes them less competitive compared with those who are actively involved in the job market. Over time, this impacts their job search process, resulting in a few more weeks needed to secure employment. We also found that higher wages reduce job search time and encourage people to find employment earlier. Higher wages may also attract individuals to rejoin the labor market and start work because there is a higher opportunity cost of being unemployed. In columns 1, 3, and 5, the results of the first-stage show that the instrumental variable, measured as the distance from the borrower to the location of the lander, is positive in all three models. Furthermore, the value of the F-statistics is high and above the critical value, indicating that the instrument is not weak (Stock & Yogo, 2005).

5.3. Analysis of Working Hours

In

Table 6. Working hours analysis—OLS and 2SLS models.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
	OLS	OLS	First Stage	Second Stage	First Stage	Second Stage
Credit Constrained	−0.001			0.802 ***
	(0.005)			(0.234)
Rejected Borrower		0.004				1.149 ***
		(0.006)				(0.373)
Distance to Lender			0.121 ***		0.0846 ***
			(0.0303)		(0.0244)
Wage	−0.072 ***	−0.072 ***	0.000884	−0.0720 ***	0.0165 ***	−0.0902 ***
	(0.003)	(0.003)	(0.00667)	(0.00625)	(0.00536)	(0.00904)
Wage Sq.	0.014 ***	0.014 ***	−0.00302 **	0.0161 ***	−0.00380 ***	0.0180 ***
	(0.001)	(0.001)	(0.00118)	(0.00129)	(0.000947)	(0.00184)
Savings	0.001 ***	0.001 ***	−0.00924 ***	0.00838 ***	−0.00611 ***	0.00799 ***
	(0.000)	(0.000)	(0.000754)	(0.00226)	(0.000607)	(0.00239)
Occupational Risk	0.028 ***	0.028 ***	−0.0227	0.0460 ***	0.000177	0.0276
	(0.009)	(0.009)	(0.0181)	(0.0177)	(0.0145)	(0.0188)
Age	0.005 ***	0.005 ***	0.00111	0.00416	0.00287	0.00175
	(0.001)	(0.001)	(0.00296)	(0.00280)	(0.00238)	(0.00327)
Age Sq.	−0.000 ***	−0.000 ***	−3.06 × ${10}^{-5}$	−2.83 × ${10}^{-5}$	−4.09 × ${10}^{-5}$	−5.81 × ${10}^{-6}$
	(0.000)	(0.000)	(3.31 × ${10}^{-5}$)	(3.19 × ${10}^{-5}$)	(2.66 × ${10}^{-5}$)	(3.78 × ${10}^{-5}$)
Auto Ownership	−0.009	−0.009	−0.0174	0.00724	0.00506	−0.0125
	(0.005)	(0.005)	(0.0111)	(0.0114)	(0.00896)	(0.0116)
Home Mortgage	0.033 ***	0.034 ***	−0.00276	0.0319	−0.0477 **	0.0845 ***
	(0.013)	(0.013)	(0.0265)	(0.0249)	(0.0213)	(0.0322)
Family Size	−0.001	−0.001	0.0183 ***	−0.0161 ***	0.00635 ***	−0.00872 **
	(0.001)	(0.001)	(0.00290)	(0.00508)	(0.00233)	(0.00384)
Female	−0.050 ***	−0.050 ***	0.00186	−0.0521 ***	−0.0296 ***	−0.0165
	(0.005)	(0.005)	(0.0109)	(0.0103)	(0.00879)	(0.0157)
Married	0.016 ***	0.016 ***	−0.0591 ***	0.0634 ***	−0.00773	0.0250 **
	(0.006)	(0.006)	(0.0115)	(0.0176)	(0.00927)	(0.0123)
Separated	0.027 ***	0.027 ***	0.0471 ***	−0.00996	0.0505 ***	−0.0302
	(0.006)	(0.006)	(0.0119)	(0.0156)	(0.00954)	(0.0224)
Widow	0.015	0.015	0.138 ***	−0.0964 **	0.151 ***	−0.160 **
	(0.013)	(0.013)	(0.0276)	(0.0417)	(0.0222)	(0.0637)
High School	0.032 ***	0.032 ***	−0.0293 *	0.0553 ***	0.0391 ***	−0.0131
	(0.008)	(0.008)	(0.0154)	(0.0159)	(0.0124)	(0.0218)
Bachelor Degree	0.015 **	0.015 **	−0.0843 ***	0.0825 ***	0.0185	−0.00632
	(0.007)	(0.007)	(0.0149)	(0.0241)	(0.0120)	(0.0170)
Post Graduate	0.042 ***	0.042 ***	−0.0997 ***	0.120 ***	0.00388	0.0360 **
	(0.008)	(0.008)	(0.0163)	(0.0274)	(0.0131)	(0.0170)
Black	−0.048 ***	−0.049 ***	0.183 ***	−0.194 ***	0.116 ***	−0.181 ***
	(0.005)	(0.005)	(0.0106)	(0.0437)	(0.00851)	(0.0445)
Hispanic/Latino	−0.027 ***	−0.027 ***	0.0336 ***	−0.0532 ***	0.0151*	−0.0436 ***
	(0.005)	(0.005)	(0.0106)	(0.0126)	(0.00851)	(0.0123)
Asian/Indian/Hawaiian	−0.060 ***	−0.060 ***	−0.000823	−0.0588 ***	0.00547	−0.0657 ***
	(0.005)	(0.005)	(0.0113)	(0.0106)	(0.00907)	(0.0119)
Constant	3.764 ***	3.764 ***	0.308 ***	3.515 ***	0.0221	3.736 ***
	(0.032)	(0.032)	(0.0649)	(0.0949)	(0.0522)	(0.0682)
Number of Observations	11,549	11,549	11,549	11,549	11,549	11,549
Wald Chi-squared				501.96 (0.00)		557.09 (0.00)
R-Squared	0.138	0.138	0.061		0.123
1-Stage F-Statistics			37.55 (0.00)		81.48 (0.00)
Endogeneity Test
Durbin Chi-Sq.				43.70 (0.00)		43.24 (0.00)
Wu-Hausman				43.79 (0.00)		43.32 (0.00)

Notes: (i) Standard errors in parentheses. (ii) *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level.

, we report the results of OLS and 2SLS estimates regarding the intensity of labor supply that is measured by the number of hours worked by the individual i. We consider only those individuals who are actively participating in the labor market and supply labor, which reduces our sample to 11,453 actively working individuals. The impact of credit constraint on working hours is negative, although not significant, in the OLS model (column 1). This suggests that constrained individuals work less than unconstrained workers. However, the credit-constrained coefficient changes its sign when we replace it with the rejected borrowers, although still insignificant. Therefore, the OLS model gives biased results, and we need to identify the causal effect of the credit constraint on working hours.

In columns (4) and (6), we control for the endogeneity problem and report the findings from the 2SLS model. The results of the second stage show the positive and significant impact of credit constraint on the working hours supplied by the individual i. This suggests that constrained individuals increase their work by one hour of work per week to mitigate the negative effect of the credit constraint. The intensity of work increases significantly more in column 6 when we replace the credit constraint with the rejected borrowers. This means that rejected borrowers or individuals turned down in credit applications increase their work by 1.14 h per week. The empirical results align with existing literature, which suggests that individuals facing liquidity or credit constraints tend to switch to higher-paying jobs or increase their working hours to offset financial shocks (Kumar & Liang, 2018). Turning to the instrument, the distance to the financial institution has an expected sign as well as significance in the first-stage equation in columns 3 and 5. Increasing the geographical distance between the borrower and the financial institution (lender) induces more credit constraints. The F-statistics in first-stage equations (columns 3 and 4) are higher than the recommended value to avoid the weak instrument problem (Stock & Yogo, 2005). The results of the endogeneity test are significant in columns 4 and 6, rejecting the null hypothesis that credit constraint and rejected borrowers are exogenous in the OLS models. Kumar and Liang (2018) used a similar framework and concluded that credit expansion leads to a reduction in working hours when bank branch deregulation is introduced. Rossi and Trucchi (2016) analyzed data from the Italian labor market and found similar results. They concluded that liquidity-constrained individuals tend to increase their working hours.

We also extend the analysis of financial imperfections by considering the impact of wages on the labor supply hours. In columns 4 and 6, the 2SLS findings show that higher wages decrease working hours. One possible explanation for this comes from the income effect of labor supply in our credit constraint model. Higher wages increase the opportunity cost of leisure, which encourages individuals to trade off labor for leisure and supply fewer hours when wages go up. We also analyze the non-linear relationship between wages and labor supply by incorporating the squared term of wages. The existing literature shows that a permanent increase in wages tends to make the income effect dominant over the substitution effect, whereas a temporary wage increase typically results in the substitution effect outweighing the income effect that support our results (Bick et al., 2022; Boppart & Krusell, 2020). Additionally, we analyze the impact of non-labor income by incorporating savings in our credit-constrained model. Our findings conclude that savings significantly increase working hours, possibly encouraging individuals to work more in order to smooth their consumption in the future and get the benefits from interest income. Occupational risk is positive and significant in the credit constraint model. This suggests that the choice to run their own business and work for self-employment encourages people to provide more labor in the presence of financial imperfections. We also control for other social, demographic, and wealth-related variables.

In

Table 7. Alternate definitions: 2SLS regression models for working hours.

	(1)	(2)	(3)	(4)
Constraint 1	2.442 *	0.610	2.066 ***	1.189 ***
	(1.357)	(0.814)	(0.653)	(0.437)
Wage	−0.0737 ***	−0.0626 ***	−0.0710 ***	−0.0651 ***
	(0.00798)	(0.0135)	(0.00677)	(0.00589)
Wage Sq.	0.0158 ***	0.0126 ***	0.0146 ***	0.0128 ***
	(0.00179)	(0.00176)	(0.00122)	(0.000999)
Savings	0.00223 **	0.00229	0.00221 ***	0.00392 ***
	(0.00112)	(0.00173)	(0.000851)	(0.00122)
Occupational Risk	0.0894 **	0.0238 **	0.0386 **	0.0557 ***
	(0.0403)	(0.0118)	(0.0186)	(0.0175)
Observations	11,549	11,549	11,549	11,549
Wald Chi-squared	17,206 (0.00)	27,529.6 (0.00)	17,868.7 (0.00)	21,540.7 (0.00)
Endogeneity Tests
Durbin Chi-Sq.	19.28 (0.00)	0.824 (0.36)	43.56 (0.00)	18.94 (0.00)
Wu-Hausman	19.28 (0.00)	0.822 (0.36)	43.64 (0.00)	18.94 (0.00)

Notes: (i)Also included age, age squared, auto ownership, home mortgage, family size, gender, categories of education and race. Standard errors in parentheses. (ii) *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level. (iii) 1-Stage F-Statistics results are significant and 1-Stage results are available. (iv) Constraint 1: Applied for credit but not given much credit (underfunded).

, we also performed a robustness analysis by examining alternative forms of credit constraint and their impact on the intensity of working hours. We consider (i) underfunded, (ii) discouraged, (iii) priced out, and (iv) rejected–revised borrowers. In Table 7, all other forms of credit constraint significantly increase working hours in the labor market by 1.18–2.4 h per week. The impact of Constraint 1 (underfunded, applied but not given much credit) is stronger compared with other forms of credit constraint. Constraint 4 (rejected–revised) borrowers are those who reapply after their first rejection/underfunded but their credit application is again rejected. Borrowers in this category significantly supply 1.18 h more labor when they are rejected again. Hence, workers increase their intensity in the labor market by working more, by adding jobs or changing their jobs (Rossi & Trucchi, 2016). Higher wages significantly decrease labor hours, confirming the income effect of the labor supply. Both non-labor income and occupational risk, measured by savings and self-employment, respectively, significantly increase working hours in the labor market. The response of other important explanatory variables is very similar to our baseline findings in Table 6. Additionally, the instrumental variable is significant in the first-stage model and F-statistic values are above the critical level, confirming the validity of the instrument.

To further check the robustness and analyze the response of credit constraint across different phases of the life cycle, we incorporate various models for different age bands and analyze the impact of credit constraint on the intensity of work. In

Table 8. Robustness Check—Working Hours with different Age bands_2SLS Results.

Variable	Age 26–35	Age 26–40	Age 26–45	Age 26–50
Credit Constrained	0.0928	0.217 **	0.756 ***	0.842 ***
	(0.108)	(0.0993)	(0.238)	(0.201)
Wage	−0.101 ***	−0.0976 ***	−0.0632 ***	−0.0749 ***
	(0.0132)	(0.00807)	(0.0114)	(0.00950)
Wage Sq.	0.0244 ***	0.0228 ***	0.0164 ***	0.0185 ***
	(0.00335)	(0.00154)	(0.00183)	(0.00178)
Savings	0.000244	0.00380 **	0.00971 ***	0.00975 ***
	(0.00131)	(0.00149)	(0.00313)	(0.00246)
Occupational Risk	0.129 ***	0.0589 *	−0.00725	0.0509 *
	(0.0470)	(0.0314)	(0.0414)	(0.0296)
Number of Observations	2365	4164	5745	7428
Wald Chi-squared	350.97 (0.00)	515.72 (0.00)	220.91 (0.00)	279.84 (0.00)
Endogeneity Tests
Durbin Chi-Sq.	0.54 (0.46)	6.34 (0.01)	44.20 (0.00)	82.63 (0.00)
Wu-Hausman	0.54 (0.46)	6.32 (0.01)	44.37 (0.00)	83.31 (0.00)

Notes: (i) Also included age, age squared, auto ownership, home mortgage, family size, gender, categories of education and race. Standard errors in parentheses. (ii) *** Significant at the 1% level, ** Significant at the 5% level, * Significant at the 10% level. (iii) 1-Stage F-Statistics results are significant and 1-Stage results are available.

, we expand the age bands for workers 35, 40, 45, and 50 years. The results for credit-constrained individuals have been positive and almost significant in all specifications of the model. Comparison of age bands highlights that working hours increase due to the impact of credit constraints. The younger age band (26–35) works fewer hours compared with the other age bands under the effect of credit constraint. This suggests that credit-constrained individuals work more hours as they enter higher age bands. Additionally, the United States job market has higher returns for experienced as well as higher age band people (age 40, 45, 50). Hence, they respond to the credit market imperfection by contributing more to labor supply. Secondly, individuals of higher age (45 or 50) generate more debt which encourages them to increase more labor supply due to credit constraints. Our findings are consistent with the work of Rossi and Trucchi (2016). The sign and significance of labor and non-labor income have been robust across all specifications of the model. Wages also show a negative impact on working hours, confirming the income effect of labor supply, but the impact is stronger for the lower age band. Additionally, the results of the endogeneity test show significant results, which confirms a robust instrument.

6. Conclusions and Policy Implication

We used the 2022 wave of the Survey of Consumer Finance (SCF) data and show that suffering from credit constraints significantly increases both the extensive and intensive margins of the labor market as measured by labor force participation and weekly work hours, respectively. We also analyze the impact of credit market imperfections on job search outcomes as measured by time to find reemployment. Our findings show that credit constraints significantly increase the time required to find employment during the job search process. We also use alternate definitions of credit constraints such as (i) rejected, (ii) underfunded, (iii) discouraged, (iv) priced-out, and (v) rejected–revised borrowers. Across various specifications, our findings show a significant and robust impact of credit imperfection on both the individual’s desire to join the labor force as well as to increase the intensity of working hours. To address the endogeneity issue, we estimate an IV-probit model which concludes that the credit constraint significantly increases the labor force participation by almost 74.3 percentage points. The impact increases when alternative forms of credit constraints are used. In our analysis, borrowers’ distance to the lender location is used as an instrumental variable in the first-stage regression that is positive and significant.

Secondly, the 2SLS findings show that credit imperfections significantly increase the intensity of work in the labor market. Individuals work almost an hour more when faced with credit constraints. The impact becomes larger when we use an alternative form of credit constraint. In addition, we introduce the impact of financial imperfection on the labor market by examining age bands. We observe that older people face more credit constraints and also work more to smooth their consumption. They start working more, find a second job, or change their job to mitigate the effects of credit constraints. Furthermore, we also control for wages, savings, wealth, and other socio-economic variables. Risky occupations, as measured by individuals’ choice to work at their own business and/or pursue self-employed opportunities, are positively related to both labor force participation and weekly work hours.

Our findings have important policy implications for households, employers, and the government. In particular, the government should implement policies within the banking sector to facilitate access to small-scale loans for households facing financial hardship. Specifically: (i) the government can enhance loan affordability by offering interest rate subsidies on small loans; (ii) it can introduce microloan programs through microfinance institutions, providing easy access to collateral-free credit; and (iii) it can promote digital and fintech-based lending solutions, which help lower administrative and operational costs for banks. These policy measures can empower households to manage short-term financial needs through small borrowing and easing pressure in the labor market. Second, it is advisable for individuals to enhance their skills during periods of unemployment, as doing so can shorten the duration of job searching and help them secure employment more quickly, particularly when facing credit constraints. Improved skills ultimately lead to better job opportunities, including higher wages and increased availability of working hours. To support this goal, the government can implement free training programs at the local level, allowing marginalized segments of the population to access quality skill-building opportunities. Additionally, expanding access to essential resources such as internet connectivity, training sessions, and workshops through public libraries or community centers can significantly improve the learning outcomes for unskilled or low-skilled workers. Lastly, employers play a crucial role in strengthening the skills of their workforce during periods of economic contraction. Instead of opting for layoffs during difficult economic periods, employers can implement targeted training initiatives to support employees in upgrading their skills and improving their abilities. Such efforts not only improve workers’ productivity and earning potential through higher hourly wages but also contribute to their financial stability, reducing their dependence on external borrowing.

Our research has several limitations that point toward potential avenues for future study. First, we did not account for the role of online lending platforms or fintech services, which have grown significantly since the 2007–2008 financial crisis. Our dataset lacks information on credit constraints arising from these newer sources. In today’s financial landscape, individuals increasingly access financial literacy resources and have more opportunities to apply for loans through online channels. This evolution suggests that online and fintech lending could offer additional credit options and potentially reduce credit constraints. Second, we were unable to control for labor force participation and working hours based on occupational categories. Individuals in certain sectors face greater credit constraints due to inability of a few sectors to generate a high wage rate. This area presents an opportunity for further investigation. Third, the nature of the business where individuals are employed may also influence credit constraints. For example, employees working for startup firms may find it difficult to increase their working hours when credit constrained. Hence, startup firms have less capacity to generate excess employment opportunities. Due to data limitations, our study did not examine startup employers that represents a valuable direction for future research.