Mission Drift or Strategic Expansion? Non-Core Lending, Risk, and Capital in US Credit Unions

Abstract

This study investigates credit unions’ expansion into non-core lending and its association with risk and financial resilience. Using US credit union call report data from 1994 to 2024, we measure the share of purchased loans, lease receivables, and loans held for sale in non-core lending. We document robust conditional, within-credit-union associations that point to a clear risk trade-off. Credit unions with higher non-core exposure grow faster in terms of loans and membership but exhibit weaker financial buffers, including lower net worth ratios and weaker economic solvency, alongside higher delinquency. Decomposition tests indicate that loans held for sale are most strongly associated with adverse buffer and asset quality patterns, while purchased loans and lease receivables display smaller and less uniform relationships. Scale interactions suggest that these associations are generally weaker for larger institutions for both membership and assets. Post-COVID estimates indicate that the baseline relationships are broadly stable, while the growth link is becoming stronger.

1. Introduction

Credit unions occupy a distinctive position in the US financial system. Organized as member-owned cooperatives, they were historically designed to provide relationship-based financial services to a well-defined membership, with an explicit commitment to serving members through local information advantages, stable share funding, and conservative lending practices. This cooperative structure has long been viewed as a stabilizing force, contributing to relatively strong asset quality and resilience during economic downturns (Smith et al. 1981; Wilcox 2005). At the same time, the operating environment of credit unions has changed markedly over the past two decades. Competitive pressure from commercial banks, FinTech lenders, and capital market-based intermediaries has intensified, while regulatory, technological, and geographic constraints on growth have gradually loosened.

In response to these pressures, many credit unions have expanded beyond traditional member-focused lending into a range of non-core activities. These include purchasing loans originated by third parties, expanding leasing programs, and originating loans intended for sale rather than long-term balance sheet retention. In this setting, a mission drift concern arises when institutions shift away from member-oriented, relationship-based, intensive lending toward growth-seeking channels that are less anchored in the cooperative model. Such strategies are often motivated by the need to maintain loan growth, improve yield, or achieve scale in markets where organic member demand is limited. However, these activities may also weaken the informational and relational foundations of the credit union business model, exposing institutions to higher credit risk and greater balance sheet volatility, and potentially undermining the financial resilience traditionally associated with the cooperative structure.

The relevance of non-core lending is evident in recent developments among large US credit unions. Institutions such as Navy Federal Credit Union and PenFed Credit Union have grown rapidly over the past decade by expanding indirect auto lending and purchased loan programs, allowing them to compete nationally despite historically narrow membership bases. Similarly, SchoolsFirst Federal Credit Union and other large occupational credit unions have increasingly relied on indirect and non-traditional lending channels to sustain growth in mature markets. Industry analysts and supervisory agencies have repeatedly noted that these growth strategies are associated with rising concentrations in purchased loans and loans originated for sale, particularly among the largest and fastest growing credit unions (National Credit Union Administration 2010).1

Periods of economic stress have further highlighted the potential risks associated with non-core lending. During the COVID-19 pandemic and the subsequent tightening of monetary policy, credit unions with greater exposure to purchased loans and indirect lending experienced sharper increases in delinquency and charge-offs relative to their peers with more relationship-oriented loan portfolios (National Credit Union Administration 2022). Trade publications and regulatory commentary have heightened supervisory attention on loan purchase programs and balance sheet growth strategies, emphasizing concerns about underwriting standards, loss recognition, and capital adequacy (Credit Union Times 2021; National Credit Union Administration 2025). These episodes underscore why non-core lending has become a central issue for both practitioners and regulators in the credit union sector.

Despite its growing importance, systematic evidence on the consequences of non-core lending in credit unions remains limited. Much of the existing literature focuses either on commercial banks or on aggregate measures of credit union diversification and loan growth, without explicitly distinguishing between relationship-based member lending and non-core activities (Frame et al. 2002; Goddard et al. 2008). Moreover, prior studies often emphasize a single outcome such as delinquency or profitability, without tracing the broader sequence through which non-core lending affects asset quality, loss realization, and capital or solvency risk (McKillop and Wilson 2011). Given that credit unions rely almost exclusively on retained earnings to build capital, understanding these dynamics is especially important.

This study addresses these gaps by examining how non-core lending influences growth, asset quality, regulatory capital, and economic solvency in US credit unions. Using detailed call report data, we construct a comprehensive measure of non-core lending based on purchased loans, lease receivables, and loans held for sale. We analyze both contemporaneous and lagged relationships to distinguish early-stage credit stress from realized losses and cumulative balance sheet effects. In addition, we decompose non-core lending into its individual components to assess whether particular activities drive the aggregate results.

Our main findings point to a clear growth risk trade-off. Credit unions with greater exposure to non-core lending experience faster loan growth and higher member growth, indicating that these activities play an important role in expansion strategies. At the same time, non-core lending is associated with higher delinquency and faster realization of credit losses through net charge-offs, consistent with weaker asset quality and reduced informational advantages. These risk effects translate into lower regulatory capital ratios and weaker economic solvency. Decomposition analyses reveal that loans held for sale contribute most strongly to risk outcomes, while purchased loans and lease receivables play distinct but complementary roles.

Taken together, our results highlight a growth and risk trade-off associated with non-core lending. Credit unions with higher non-core exposure expand more quickly in terms of loans and membership, but they also exhibit weaker asset quality and thinner financial buffers, with the most pronounced associations concentrated in loans held for sale. These findings have direct policy implications for risk oversight. In particular, they suggest that supervisory monitoring should not rely primarily on growth metrics but explicitly incorporate balance sheet composition and the channels through which credit risk is accumulated and absorbed. A practical implication is that elevated non-core shares, especially originate-for-sale activity, can serve as a supervisory flag that warrants closer review of underwriting and due diligence practices, particularly among smaller institutions. This perspective aligns with the broader emphasis in supervisory guidance on credit risk management and capital adequacy as core elements of financial resilience (Foos et al. 2010).2

2. Literature Review

2.1. Credit Unions, Cooperative Structure, and Risk Taking

A large part of the finance literature emphasizes that credit unions differ fundamentally from commercial banks because of their cooperative ownership structure and member-based objectives. Early theoretical work argues that the absence of external shareholders reduces incentives for excessive risk taking and encourages relationship-oriented lending built on local information and repeated interactions with members (Smith et al. 1981; McKillop and Wilson 2011). Empirical studies generally support this view, showing that credit unions tend to exhibit lower volatility, more stable funding, and different risk profiles relative to investor-owned banks, particularly during periods of financial stress (Wilcox 2005; Bauer 2008; McKillop and Wilson 2011).

At the same time, the cooperative model imposes constraints. Credit unions rely almost exclusively on retained earnings to build capital, which limits their ability to absorb sustained losses or fund rapid expansion (McKillop and Wilson 2011; Goddard et al. 2009). As a result, growth-oriented strategies may require trade-offs between scale, profitability, and financial resilience. Several studies document that larger and faster growing credit unions increasingly resemble banks in their balance sheet composition and product offerings, suggesting a gradual convergence of business models over time (Frame et al. 2002; Goddard et al. 2008; Walter 2006).

This convergence has motivated a growing literature on mission drift in cooperative financial institutions. Mission drift refers to the tendency of organizations originally created to serve a narrow constituency to expand activities in ways that dilute their original purpose. In the credit union context, mission drift is often associated with geographic expansion, broader fields of membership, and greater reliance on transactional lending channels (Emmons and Schmid 1999; Leggett and Strand 2002). While such shifts may improve competitiveness, they may also erode the informational advantages that underpin traditional credit union lending.

2.2. Loan Growth, Diversification, and Asset Quality

Related studies examined the relationship between loan growth, diversification, and asset quality. In banking studies, rapid loan growth is frequently linked to subsequent increases in delinquency and charge-offs, reflecting weaker underwriting standards or adverse selection during expansionary phases (Foos et al. 2010). Research on securitization and originate-to-distribute models further showed that loans originated with the intent to sell or transfer often perform worse than loans retained on the balance sheet, in part because screening incentives are weakened (Keys et al. 2010; Purnanandam 2011). Evidence for credit unions points in a similar direction, though the mechanisms differ. Studies focusing on credit unions found that aggressive growth strategies are associated with higher credit risk and greater earnings volatility, especially when growth is driven by non-traditional lending activities rather than core member demand (Wheelock and Wilson 2011). Related evidence also linked diversification and greater reliance on non-traditional revenue channels to changes in risk profiles in the credit union sector (Esho et al. 2005). However, most of this literature relied on broad measures of loan growth or diversification and did not explicitly distinguish between relationship-based lending and non-core activities such as purchased loans or loans held for sale. Moreover, asset quality is often treated as a static outcome rather than a dynamic process. Delinquency and charge-offs are frequently analyzed separately, even though they represent different stages of credit risk realization. This limits our understanding of how risk emerges, is managed, and ultimately affects financial stability in credit unions.

2.3. Capital, Solvency, and Loss Absorption in Credit Unions

Capital adequacy is central to the stability of credit unions because of their limited access to external capital markets and their reliance on retained earnings for capital formation (McKillop and Wilson 2011; Goddard et al. 2009). Prior research showed that declines in asset quality translate into lower net worth ratios through reduced earnings and higher provisioning, constraining future lending and growth (Goddard et al. 2014). Regulatory frameworks such as prompt corrective action explicitly link capital ratios to supervisory intervention, reinforcing the importance of capital buffers. Beyond regulatory capital, several studies highlighted the importance of economic solvency measures that capture balance sheet resilience more broadly. In banking research, economic capital and solvency indicators often provide early warning signals of distress even when regulatory capital ratios remain above minimum thresholds (Berger and Bouwman 2013). For credit unions, similar concerns arise when asset composition shifts toward riskier or less liquid exposures, even if the reported capital ratios appear adequate (Goddard et al. 2009). Despite this, empirical work on credit union solvency remains relatively sparse. Existing studies typically focused on net worth ratios and did not examine alternative measures of balance sheet strength that account for asset quality and valuation adjustments. This leaves open the question of how non-core lending affects not only regulatory capital, but also underlying economic solvency.

2.4. Contribution of This Study

This study contributes to the literature in several focused ways. First, it provides systematic evidence on how non-core lending activities are associated with growth, asset quality, and financial resilience in credit unions. Rather than emphasizing overall loan growth or diversification, the analysis distinguishes between traditional member-based lending and non-core activities such as purchased loans, lease receivables, and loans held for sale, which are increasingly relevant in practice. Second, the study highlights the dynamic nature of credit risk in credit unions by separating early-stage indicators of distress from realized losses and longer-term balance sheet effects. By examining delinquency, net charge-offs, capital ratios, and economic solvency measures within a unified framework, this study clarifies how non-core lending affects different stages of risk realization. Third, the decomposition results provide insight into which types of non-core lending are most strongly associated with risk and capital outcomes. This allows for a more nuanced interpretation of aggregate effects and helps link empirical findings to supervisory and policy concerns. Overall, the paper adds to the growing literature on mission drift and risk taking in cooperative financial institutions by documenting a consistent growth and risk trade-off associated with non-core lending in credit unions.

3. Methodology

3.1. Data and Sample Construction

Our raw sample consisted of US credit union call report data covering a large panel of institutions from 1994 to 2024.3 The data includes detailed information on the balance sheets, income statements, asset quality, and membership characteristics reported to regulators. The reporting framework is uniform across institutions and over time, which allows for consistent measurement of financial variables and institutional behavior over a long sampling period.

Observations with missing values for key variables were excluded to ensure comparability across specifications. Our final sample consisted of 156,570 credit union year observations. To limit the influence of extreme observations, all continuous variables were winsorized at the 1st and 99th percentiles using Stata 17’s winsor command. This procedure reduces the impact of outliers while preserving the underlying cross-sectional and time-series variation in the data. The resulting panel is unbalanced, reflecting entry, exit, and consolidation within the credit union sector over the years.

3.2. Key Variable Construction

The primary independent variable was non-core lending, which captures lending activities that deviate from traditional member-based relationship lending. We measured non-core lending as the ratio of purchased loans, lease receivables, and loans held for sale to total loans. Purchased loans reflect credit exposures acquired from external originators, lease receivables represent non-traditional lending arrangements often associated with commercial activity, and loans held for sale capture originate-to-distribute behavior. Together, these components are proxies for lending that relies less on member relationships and local information advantages. To examine the underlying channels, we also constructed each component separately and used them in decomposition regressions. This allowed us to assess whether specific non-core activities drive the aggregate effects.

We considered several categories of dependent variables. Asset quality was measured using delinquent loans scaled by total net worth and net charge-offs scaled by loans outstanding. Delinquency captures early-stage credit distress, while net charge-offs reflect realized credit losses. Capital risk is measured using net worth to total assets, which is the standard regulatory capital ratio for credit unions, and delinquent loans are measured relative to net worth, which captures capital vulnerability. An adjusted solvency evaluation ratio that reflects the extent to which asset values back member claims after accounting for reporting adjustments over time was used as a proxy for economic solvency. This measure was constructed from a small set of balance sheet items to capture the extent to which adjusted asset values back member shares and deposits.4 Growth outcomes included loan growth, asset growth, and member growth, which were measured as annualized changes.

We also included a comprehensive set of control variables to isolate the effect of non-core lending. Size and membership scale were controlled for using the logarithm of total assets and the logarithm of members, capturing economies of scale and diversification effects. Credit union age was included to account for lifecycle differences between younger and more established institutions. Productivity and operating structure were controlled for using measures of members per full time equivalent employee, and land and building assets. Profitability and cost structure were controlled for using return on assets and operating expense ratio. These controls capture differences in business models, efficiency, and financial capacity that may influence both non-core lending decisions and outcomes. Detailed definitions of all these variables are given in Appendix A.

All regressions included a comprehensive set of fixed effects. Year fixed effects control for macroeconomic conditions, regulatory changes, and industry wide shocks. Data system fixed effects account for reporting system differences. Credit union type fixed effects control for institutional characteristics tied to the charter or field of membership. Regional fixed effects absorb persistent geographic differences in economic conditions. Peer group fixed effects control for size-based supervisory classifications. Together, these fixed effects substantially reduce concerns about omitted variable bias. However, we do acknowledge the presence of endogeneity and limited causal inference under this setup, and our results should be interpreted as robust conditional associations. The contribution therefore rests on the consistency of the evidence across alternative specifications and robustness checks

3.3. Empirical Specification

We estimate a general baseline panel regression with the following form:

$$Y_{i,t}=\alpha +\beta {\mathit{\text{Non-core}} \mathit{Loans}}_{i,t-1}+\gamma X_{i,t-1}+{\mu }_i+{\lambda }_t+{\epsilon }_{i,t}$$

(1)

where Y_i,t denotes an outcome variable for credit union i in year t, including measures of asset quality, capital risk, solvency, or growth. Non-core Loans is the ratio of total non-core loans to total loans. X is a vector of control variables.5 μ_i and λ_t represent the full set of institution-related and time-related fixed effects described above. Standard errors of all our regressions were clustered by credit union to account for within-credit-union serial correlation and heteroscedasticity over time, which yields more reliable inference in a panel setting.6

3.4. Decomposition of Non-Core Lending

To examine whether specific non-core activities drive the aggregate results, we estimated decomposition regressions that replace Non-core Loans with its individual components:

$$Y_{i,t}=\alpha +{\beta }_1{\mathit{Purchased} \mathit{Loans}}_{i,t-1}+{\beta }_2{\mathit{Lease} \mathit{Receivables}}_{i,t-1}+{\beta }_3{\mathit{Loans} \mathit{Held} \mathit{for} \mathit{Sale}}_{i,t-1}+\gamma X_{i,t-1}+{\mu }_i+{\lambda }_t+{\epsilon }_{i,t}$$

(2)

Each component was scaled by total loans and entered simultaneously to capture their marginal associations with the outcome variables. This specification allowed us to assess whether certain types of non-core lending are more strongly linked to asset quality deterioration, capital risk, or growth. The same controls and fixed effects were included to ensure comparability with the baseline regressions.

4. Empirical Results and Discussion

4.1. Descriptive Statistics

Table 1. Descriptive statistics. This table reports the descriptive statistics for the main variables in the analysis using 156,570 credit union-year observations. For each variable, the table presents the mean, standard deviation, minimum, 25th percentile, median, 75th percentile, and maximum. Variables are defined in Appendix A.

Variables (n = 156,570)	Mean	S.D.	Min	0.25	Median	0.75	Max
Non-Core Loans	0.001	0.006	0.000	0.000	0.000	0.000	0.044
Purchased Loans	0.000	0.002	0.000	0.000	0.000	0.000	0.033
Lease Receivables	0.000	0.002	0.000	0.000	0.000	0.000	0.021
Loan Held for Sale	0.000	0.002	0.000	0.000	0.000	0.000	0.017
Net worth	0.136	0.060	0.058	0.096	0.119	0.157	0.385
Delinquent Loans	0.079	0.119	0.000	0.016	0.041	0.091	0.758
Solvency	1.165	0.095	1.057	1.106	1.136	1.188	1.631
Net Charge-Offs	0.073	0.117	−0.080	0.011	0.040	0.088	0.755
Loan Growth	0.376	1.616	−3.810	−0.557	0.294	1.172	6.619
Asset Growth	0.444	1.037	−2.251	−0.161	0.376	0.963	4.290
Member Growth	0.006	0.859	−2.833	−0.350	−0.021	0.311	4.299
Member Size	8.091	1.597	4.625	6.987	7.990	9.145	11.939
Size	16.803	2.020	11.656	15.489	16.786	18.137	21.403
Age	4.019	0.486	0.000	3.850	4.043	4.234	7.613
Land and Building	0.013	0.016	0.000	0.000	0.006	0.022	0.066
ROA	0.050	0.103	−0.466	0.013	0.057	0.104	0.325
Operating Expenses	0.445	0.173	0.113	0.330	0.429	0.537	1.113

reports the summary statistics for the main variables. A key feature of the data is that non-core lending is economically meaningful but highly concentrated. The typical credit union reports very limited exposure, yet the distribution has a long right tail, indicating that a subset of institutions relies more heavily on purchased loans, lease receivables, or loans held for sale. This type of dispersion is consistent with prior evidence that US credit unions differ substantially in business model choices and diversification strategies (Goddard et al. 2008).

Measures of financial condition and performance also display wide variation. Asset quality indicators such as delinquency and net charge-offs are low on average but vary considerably across institutions and over time, which potentially aligns with evidence that faster expansion and portfolio shifts can be linked to subsequent risk outcomes (Foos et al. 2010). Capital buffers and solvency measures similarly vary across credit unions, reflecting differences in balance sheet choices and loss absorption capacity.

Growth outcomes exhibit especially wide ranges, with both positive and negative realizations for loan, asset, and membership growth, reflecting expansion, contraction, and consolidation dynamics in the sector. The presence of meaningful variation in membership size and asset size is also important for our subsequent heterogeneity analysis and aligns with the evidence of scale effects among credit unions (Wheelock and Wilson 2011). Finally, our sample includes credit unions with different fields of membership structures and growth opportunities, which prior research suggests can be associated with differences in risk taking and performance (Frame et al. 2002).

4.2. Correlation Analysis

Table 2. Correlation Matrix. This table reports the pairwise Pearson correlations between the main variables. Correlations are shown for the aggregate non-core loan share, its components, dependent variables, and controls. Variables are defined in Appendix A. * indicates statistical significance at the 5 percent level.

		(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)	(16)	(17)
(1)	Non-core Loans	1.0000
(2)	Purchased Loans	0.3841 *	1.0000
(3)	Lease Receivables	0.6467 *	0.0242 *	1.0000
(4)	Loan Held for Sale	0.5950 *	−0.0118 *	0.0231 *	1.0000
(5)	Net worth	−0.0777 *	−0.0116 *	−0.0386 *	−0.1007 *	1.0000
(6)	Delinquent Loans	0.0034	0.0180 *	0.0025	−0.0160 *	−0.1613 *	1.0000
(7)	Solvency	−0.0691 *	−0.0100 *	−0.0349 *	−0.0898 *	0.9864 *	−0.1362 *	1.0000
(8)	Net Charge-offs	−0.0077 *	0.0026	0.0066 *	−0.0264 *	0.0178 *	0.2987 *	0.0302 *	1.0000
(9)	Loan Growth	0.0418 *	0.0019	0.0086 *	0.0697 *	−0.1237 *	−0.0927 *	−0.1198 *	−0.2402 *	1.0000
(10)	Asset Growth	0.0627 *	0.0337 *	0.0179 *	0.0844 *	−0.2207 *	−0.0933 *	−0.2047 *	−0.1626 *	0.3384 *	1.0000
(11)	Member Growth	0.0510 *	0.0102 *	0.0180 *	0.0757 *	−0.1163 *	0.0053 *	−0.1110 *	−0.0750 *	0.3395 *	0.3815 *	1.0000
(12)	Member Size	0.1676 *	−0.0038	0.0861 *	0.2454 *	−0.4093 *	−0.1501 *	−0.4018 *	−0.0563 *	0.1868 *	0.2394 *	0.1725 *	1.0000
(13)	Size	0.1510 *	−0.0145 *	0.0720 *	0.2356 *	−0.3920 *	−0.2159 *	−0.3904 *	−0.1159 *	0.1903 *	0.2202 *	0.1779 *	0.9545 *	1.0000
(14)	Age	0.0052 *	−0.0270 *	−0.0058 *	0.0408 *	0.0150 *	−0.1232 *	0.0065 *	−0.0735 *	−0.0211 *	−0.0348 *	−0.0541 *	0.1613 *	0.2181 *	1.0000
(15)	Land and Building	0.0457 *	−0.0106 *	0.0133 *	0.0908 *	−0.2409 *	−0.0337 *	−0.2302 *	−0.0119 *	0.0961 *	0.1294 *	0.1077 *	0.4274 *	0.4003 *	0.0929 *	1.0000
(16)	ROA	0.0428 *	0.0166 *	0.0143 *	0.0623 *	0.0510 *	−0.1403 *	0.0382 *	−0.1630 *	0.1531 *	0.2736 *	0.1721 *	0.1943 *	0.1975 *	−0.0390 *	0.0226 *	1.0000
(17)	Operating Expenses	0.0114 *	−0.0010	0.0004	0.0147 *	−0.0627 *	0.2286 *	−0.0377 *	0.2266 *	−0.0408 *	0.0017	−0.0214 *	0.0235 *	−0.1279 *	−0.1000 *	0.2140 *	−0.2450 *	1.0000

reports the pair-wise correlations between the main variables. The correlation patterns are broadly consistent with the later regression results and do not indicate concerns of collinearity. Non-core loans are positively related to the three non-core components by construction and are modestly positively correlated with growth outcomes, particularly loans and member growth. Non-core lending is also negatively correlated with net worth and the solvency measure, suggesting that a greater reliance on non-traditional lending is associated with thinner buffers. Asset quality measures generally exhibit expected correlations. Delinquency is positively correlated with net charge-offs, and both are negatively related to capital and solvency. Finally, the correlations between purchased loans, lease receivables, and loans held for sale are positive but not sufficiently high to prevent joint inclusion in the decomposition empirical specifications.7

4.3. Non-Core Lending, Asset Quality, and Capital Risk

Table 3. Non-core loans, capital, solvency, and asset quality. This table reports the panel regression results for non-core loans and (1) net worth, (2) delinquent loans, (3) solvency, and (4) net charge-offs. All specifications include the stated control variables and fixed-effects for year, data system, credit union type, region, and peer group. Standard errors of all specifications are clustered by credit union to account for within-credit-union serial correlation and heteroscedasticity over time. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Variables are defined in Appendix A.

	(1)	(2)	(3)	(4)
	Net worth	Delinquent Loans	Solvency	Net Charge-offs
Non-Core Loans	−0.1498 ***	0.2817 ***	−0.1482 ***	−0.1166 **
	(0.032)	(0.081)	(0.047)	(0.058)
Member Size	−0.0169 ***	0.0120 ***	−0.0272 ***	0.0204 ***
	(0.002)	(0.002)	(0.003)	(0.002)
Size	0.0008	−0.0169 ***	0.0003	−0.0123 ***
	(0.002)	(0.003)	(0.003)	(0.002)
Age	0.0143 ***	−0.0127 ***	0.0213 ***	−0.0062 ***
	(0.001)	(0.002)	(0.002)	(0.002)
Land and Building	−0.2010 ***	0.1185 ***	−0.2884 ***	−0.1610 ***
	(0.030)	(0.044)	(0.047)	(0.034)
ROA	0.0789 ***	−0.0826 ***	0.1185 ***	−0.1291 ***
	(0.004)	(0.006)	(0.007)	(0.007)
Operating Expenses	−0.0004	0.1064 ***	0.0119 *	0.0995 ***
	(0.004)	(0.007)	(0.007)	(0.005)
Constant	0.2015 ***	0.2716 ***	1.2874 ***	0.1039 ***
	(0.019)	(0.032)	(0.032)	(0.022)
Year FEs	Yes	Yes	Yes	Yes
Data System FEs	Yes	Yes	Yes	Yes
CU Type FEs	Yes	Yes	Yes	Yes
Region FEs	Yes	Yes	Yes	Yes
Peer Group FEs	Yes	Yes	Yes	Yes
Adjusted R2	0.24	0.13	0.23	0.09
Observations	156,570	156,570	156,570	156,570

reports the baseline regressions linking the aggregate non-core loan share to capital, asset quality, and solvency outcomes. Across specifications, non-core lending is associated with weaker buffers and poorer loan performance. The coefficient for non-core loans is negative and highly statistically significant in the net worth regression, indicating that credit unions with greater non-core exposure tend to have lower net worth ratios. A similar negative and statistically significant relationship is observed for the solvency measure, suggesting that non-core lending is also associated with weaker balance sheet resilience. On the asset quality side, non-core loans are positively and significantly related to delinquent loans, implying higher credit stress among institutions relying more heavily on non-core activities. The net charge-offs regression also shows a negative and statistically significant coefficient, indicating that higher non-core exposure is associated with lower net charge-offs in the same period, which is consistent with the idea that realized loss recognition may not move one-for-one with contemporaneous delinquency and can reflect timing, portfolio turnover, or loan management practices rather than only underlying credit deterioration. In particular, moves into loans held for sale and subsequent sales can affect the timing of loss recognition and recoveries so net charge-offs may capture both realized credit losses and shifts in recognition timing due to sale and workout decisions.

The control variables generally show expected signs. Higher profitability is associated with stronger net worth and solvency and with lower delinquency and charge-offs, consistent with retained earnings supporting buffers. Larger and older credit unions exhibit slower deterioration in some dimensions but also show systematic differences in capital and loss measures, reflecting heterogeneity in business models. Operating expenses load positively on delinquency and charge-offs, consistent with cost inefficiency being correlated with weaker loan performance. Finally, the inclusion of year and multiple institutional fixed effects is important because it absorbs common macroeconomic conditions and persistent differences across credit unions, allowing the coefficients on non-core loans to be interpreted as within-credit-union associations over time.

4.4. Non-Core Lending and Growth Outcomes

We then examine the relationship between non-core lending and growth outcomes and the results are shown in

Table 4. Non-core loans and growth. This table reports the panel regression results for non-core loans and (1) loan growth, (2) asset growth, and (3) member growth. All specifications include the stated control variables and fixed effects for year, data system, credit union type, region, and peer group. Standard errors of all specifications are clustered by credit union to account for within-credit-union serial correlation and heteroscedasticity over time. *** indicates statistical significance at the 1 percent levels. Variables are defined in Appendix A.

	(1)	(2)	(3)
	Loan Growth	Asset Growth	Member Growth
Non-Core Loans	3.2643 ***	0.5653	1.5935 ***
	(0.828)	(0.521)	(0.507)
Member Size	0.0112	0.0674 ***	−0.1717 ***
	(0.016)	(0.010)	(0.012)
Size	−0.1035 ***	−0.2243 ***	0.0712 ***
	(0.016)	(0.010)	(0.010)
Age	−0.2375 ***	−0.1245 ***	−0.1629 ***
	(0.021)	(0.012)	(0.014)
Land and Building	3.4923 ***	2.1904 ***	3.3466 ***
	(0.330)	(0.222)	(0.232)
ROA	1.7376 ***	2.5074 ***	1.2409 ***
	(0.063)	(0.040)	(0.035)
Operating Expenses	0.1229 ***	0.2653 ***	0.2880 ***
	(0.043)	(0.027)	(0.027)
Constant	2.7882 ***	3.8957 ***	0.6181 ***
	(0.202)	(0.125)	(0.119)
Year FEs	Yes	Yes	Yes
Data System FEs	Yes	Yes	Yes
CU Type FEs	Yes	Yes	Yes
Region FEs	Yes	Yes	Yes
Peer Group FEs	Yes	Yes	Yes
Adjusted R2	0.10	0.27	0.08
Observations	156,570	156,570	156,570

. The coefficient on non-core loans is positive and statistically significant for both loan growth and member growth, indicating that credit unions with greater non-core exposure tend to expand their loan portfolios and membership more rapidly. In contrast, the association with asset growth is positive but not statistically significant, suggesting that non-core lending is more closely linked to growth in lending activity and membership than to broad balance sheet expansion. One interpretation is that credit unions may reallocate within the balance sheet as they pursue non-core lending or face funding and capital constraints that limit overall asset expansion even when lending grows.

The coefficients of the control variables remain intuitive. Larger credit unions and older credit unions grow more slowly in terms of loans and assets, consistent with the maturity and scale effects documented in the credit union literature (Wheelock and Wilson 2011). ROA, which serves as a proxy for profitability, is strongly positively associated with all growth measures, consistent with stronger earnings supporting expansion capacity. The positive coefficients on operating expenses are also consistent with a higher operating intensity accompanying periods of growth. Taken together, these results complement the broader theme of the paper by showing that non-core lending is associated with faster expansion along key dimensions, a pattern that aligns with evidence linking growth strategies to shifts in portfolio composition and risk taking in financial intermediaries (Foos et al. 2010; Goddard et al. 2008).

4.5. Decomposition of Non-Core Lending

To better understand what drives the aggregate non-core lending results,

Table 5. Components of non-core lending, capital, solvency, and asset quality. This table decomposes non-core lending into purchased loans, lease receivables, and loans held for sale, and regress these components to (1) net worth, (2) delinquent loans, (3) solvency, and (4) net charge-offs. All regressions include the stated control variables and fixed effects for year, data system, credit union type, region, and peer group. Standard errors of all specifications are clustered by credit union to account for within-credit-union serial correlation and heteroscedasticity over time. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Variables are defined in Appendix A.

	(1)	(2)	(3)	(4)
	Net Worth	Delinquent Loans	Solvency	Net Charge-offs
Purchased Loans	−0.1851 **	0.5785 **	−0.2532 **	−0.0435
	(0.072)	(0.237)	(0.111)	(0.184)
Lease Receivables	−0.1107	0.2449	−0.0691	0.2233
	(0.082)	(0.212)	(0.120)	(0.154)
Loan Held for Sale	−0.4670 ***	0.5602 ***	−0.4576 ***	−0.8853 ***
	(0.083)	(0.190)	(0.119)	(0.120)
Member Size	−0.0169 ***	0.0121 ***	−0.0273 ***	0.0203 ***
	(0.002)	(0.002)	(0.003)	(0.002)
Size	0.0008	−0.0169 ***	0.0003	−0.0121 ***
	(0.002)	(0.003)	(0.003)	(0.002)
Age	0.0144 ***	−0.0127 ***	0.0213 ***	−0.0061 ***
	(0.001)	(0.002)	(0.002)	(0.002)
Land and Building	−0.2006 ***	0.1177 ***	−0.2879 ***	−0.1601 ***
	(0.030)	(0.044)	(0.047)	(0.034)
ROA	0.0791 ***	−0.0828 ***	0.1188 ***	−0.1284 ***
	(0.004)	(0.006)	(0.007)	(0.007)
Operating Expenses	−0.0002	0.1063 ***	0.0121 *	0.1000 ***
	(0.004)	(0.007)	(0.007)	(0.005)
Constant	0.2007 ***	0.2720 ***	1.2868 ***	0.1021 ***
	(0.019)	(0.032)	(0.032)	(0.022)
Year FEs	Yes	Yes	Yes	Yes
Data System FEs	Yes	Yes	Yes	Yes
CU Type FEs	Yes	Yes	Yes	Yes
Region FEs	Yes	Yes	Yes	Yes
Peer Group FEs	Yes	Yes	Yes	Yes
Adjusted R2	0.24	0.13	0.23	0.09
Observations	156,570	156,570	156,570	156,570

decomposes total non-core loans into its three components (purchased loans, lease receivables, and loans held for sale) and re-estimates the baseline specifications. The estimates show that the aggregate relationships mask meaningful differences across activities. Loans held for sale emerge as the most risk-intensive component. It is strongly associated with lower net worth and lower solvency, and with higher delinquency and markedly higher net charge-offs. Purchased loans are also linked to weaker buffers and higher delinquency, although the net charge-offs estimate is not statistically significant. Lease receivables display a weaker association with capital buffer measures but are positively related to delinquency and net charge-offs, indicating that this component is also tied to measurable credit risk. Overall, these results align with prior evidence that diversification and growth strategies can be associated with risk outcomes, and that the composition of lending matters for performance and resilience (Goddard et al. 2008; Wheelock and Wilson 2011).

Table 6. Components of non-core lending and growth. This table decomposes non-core lending into purchased loans, lease receivables, and loans held for sale, and regress these components to (1) loan growth, (2) asset growth, and (3) member growth. All regressions include the stated control variables and fixed effects for year, data system, credit union type, region, and peer group. Standard errors of all specifications are clustered by credit union to account for within-credit-union serial correlation and heteroscedasticity over time. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Variables are defined in Appendix A.

	(1)	(2)	(3)
	Loan Growth	Asset Growth	Member Growth
Purchased Loans	5.3223 *	1.7067	0.7580
	(2.926)	(1.749)	(1.520)
Lease Receivables	−4.2253 **	−2.7451 **	0.2469
	(1.867)	(1.207)	(1.201)
Loan Held for Sale	16.0990 ***	6.5454 ***	7.9244 ***
	(1.885)	(1.298)	(1.412)
Member Size	0.0124	0.0679 ***	−0.1715 ***
	(0.016)	(0.010)	(0.012)
Size	−0.1053 ***	−0.2253 ***	0.0702 ***
	(0.016)	(0.010)	(0.010)
Age	−0.2381 ***	−0.1247 ***	−0.1632 ***
	(0.021)	(0.012)	(0.014)
Land and Building	3.4702 ***	2.1831 ***	3.3409 ***
	(0.330)	(0.222)	(0.231)
ROA	1.7247 ***	2.5015 ***	1.2355 ***
	(0.063)	(0.040)	(0.035)
Operating Expenses	0.1146 ***	0.2610 ***	0.2839 ***
	(0.043)	(0.027)	(0.027)
Constant	2.8136 ***	3.9094 ***	0.6340 ***
	(0.202)	(0.125)	(0.119)
Year FEs	Yes	Yes	Yes
Data System FEs	Yes	Yes	Yes
CU Type FEs	Yes	Yes	Yes
Region FEs	Yes	Yes	Yes
Peer Group FEs	Yes	Yes	Yes
Adjusted R2	0.10	0.27	0.08
Observations	156,570	156,570	156,570

reports the growth regressions for the decomposition of non-core loans. The results make clear that the positive association between non-core lending and growth is not evenly shared across components. Loans held for sale are the dominant driver. Its coefficients are large and highly statistically significant across all three dependent variables, indicating that credit unions with greater reliance on loans originated for sale expand faster in terms of loans, total assets, and membership. In other words, the originate-for-sale channel appears tightly linked to expansion behavior rather than being a minor balance sheet item.

Purchased loans exhibit a more limited growth footprint. The coefficient is positive and statistically significant for loan growth, but not for asset growth or member growth. This pattern is consistent with purchased loans being used primarily to supplement loan volumes rather than to scale up the overall institution. By contrast, lease receivables are negatively related to loan growth and asset growth, with statistically significant estimates, suggesting that greater exposure to leasing is associated with slower expansion in the broader balance sheet. One plausible interpretation is that lease receivables are a proxy for more niche lending activity or for portfolio substitution away from conventional lending growth. Overall, the decomposition results complement our main findings by showing that the growth link attributed to non-core lending is concentrated in loans held for sale, consistent with the evidence from banking that originate-to-distribute activity is associated with faster credit expansion and different incentive structures relative to relationship lending (Keys et al. 2010; Purnanandam 2011).

4.6. Scale Effects and Non-Core Lending

To assess whether the non-core lending associations documented above are concentrated among smaller credit unions or persist across the size distribution,

Table 7. Scale interactions. This table tests whether the relationship between non-core lending and outcomes varies with institutional scales. Columns (1) to (4) include an interaction between Non-core Loans and Member Size, with dependent variables Net Worth, Delinquent Loans, Solvency, and Net Charge-offs. Columns (5) to (8) repeat the analysis using an interaction between Non-core Loans and Size. All regressions include the stated controls and fixed-effects for year, data system, credit union type, region, and peer group. Standard errors of all specifications are clustered by credit union to account for within-credit-union serial correlation and heteroscedasticity over time. *, **, and *** indicate statistical significance at the 10 percent, 5 percent, and 1 percent levels, respectively. Variables are defined in Appendix A.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Net Worth	Delinquent Loans	Solvency	Net Charge-Offs	Net Worth	Delinquent Loans	Solvency	Net Charge-Offs
Non-Core Loans × Member Size	0.0485 **	−0.1071 **	0.1004 ***	−0.0696 *
	(0.019)	(0.048)	(0.034)	(0.040)
Non-Core Loans × Size					0.0293 *	−0.0942 **	0.0656 **	−0.0568 *
					(0.015)	(0.042)	(0.026)	(0.034)
Non-Core Loans	−0.5910 ***	1.2563 ***	−1.0614 ***	0.5166	−0.6748 **	1.9687 **	−1.3237 ***	0.9002
	(0.187)	(0.479)	(0.330)	(0.391)	(0.280)	(0.791)	(0.478)	(0.627)
Member Size	−0.0170 ***	0.0122 ***	−0.0274 ***	0.0205 ***	−0.0169 ***	0.0121 ***	−0.0273 ***	0.0204 ***
	(0.002)	(0.002)	(0.003)	(0.002)	(0.002)	(0.002)	(0.003)	(0.002)
Size	0.0008	−0.0169 ***	0.0003	−0.0123 ***	0.0007	−0.0168 ***	0.0002	−0.0122 ***
	(0.002)	(0.003)	(0.003)	(0.002)	(0.002)	(0.003)	(0.003)	(0.002)
Age	0.0143 ***	−0.0127 ***	0.0213 ***	−0.0061 ***	0.0143 ***	−0.0127 ***	0.0213 ***	−0.0061 ***
	(0.001)	(0.002)	(0.002)	(0.002)	(0.001)	(0.002)	(0.002)	(0.002)
Land and Building	−0.2005 ***	0.1173 ***	−0.2873 ***	−0.1617 ***	−0.2007 ***	0.1174 ***	−0.2877 ***	−0.1616 ***
	(0.030)	(0.044)	(0.047)	(0.034)	(0.030)	(0.044)	(0.047)	(0.034)
ROA	0.0789 ***	−0.0826 ***	0.1185 ***	−0.1291 ***	0.0789 ***	−0.0825 ***	0.1185 ***	−0.1291 ***
	(0.004)	(0.006)	(0.007)	(0.007)	(0.004)	(0.006)	(0.007)	(0.007)
Operating Expenses	−0.0003	0.1064 ***	0.0119 *	0.0995 ***	−0.0004	0.1064 ***	0.0119 *	0.0995 ***
	(0.004)	(0.007)	(0.007)	(0.005)	(0.004)	(0.007)	(0.007)	(0.005)
Constant	0.2021 ***	0.2702 ***	1.2888 ***	0.1030 ***	0.2021 ***	0.2697 ***	1.2888 ***	0.1027 ***
	(0.019)	(0.032)	(0.032)	(0.022)	(0.019)	(0.033)	(0.032)	(0.022)
Year FEs	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Data System FEs	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
CU Type FEs	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Region FEs	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Peer Group FEs	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Adjusted R2	0.24	0.13	0.23	0.09	0.24	0.13	0.23	0.09
Observations	156,570	156,570	156,570	156,570	156,570	156,570	156,570	156,570

shows the interaction terms between non-core loans and institutional size scale measures. Columns (1) to (4) use member size as the scaling variable. The interaction is positive for net worth and solvency and negative for delinquency and net charge-offs. Together with the negative level effect of non-core loans on net worth and their positive association with delinquency, these signs imply that non-core lending is linked to weaker capital buffers and poorer asset quality on average, but the magnitude of these relationships is moderate and become smaller for credit unions with larger memberships. This pattern is consistent with scale-related advantages such as broader diversification and more developed underwriting and monitoring infrastructure, which can partially dampen the risk consequences of non-core activities (Goddard et al. 2008; Wheelock and Wilson 2011).

Columns (5) to (8) repeat this interaction analysis using instead Size measured by total assets. The results are qualitatively similar. The interaction term is positive for net worth and solvency and negative for delinquency and net charge-offs, indicating that the adverse associations tied to non-core lending are attenuated for larger balance sheets. In other words, while non-core exposure is linked to weaker capital buffers and poorer credit performance on average, larger credit unions appear less sensitive to these effects, consistent with scale related benefits in diversification, internal controls, and risk management capacity that are commonly emphasized for financial intermediaries (Berger and Bouwman 2013; Foos et al. 2010).

4.7. Impact of COVID-19

To assess whether the relationship between non-core lending and credit union outcomes changes after the COVID period, we added an interaction between the post-COVID indicator (i.e., a dummy equals to 1 if the year is 2020 and later) and non-core loans to the baseline specifications while keeping the same controls and fixed effects, including year fixed effects, and the results are shown in

Table 8. Post-COVID interaction, capital, solvency, and asset quality. This table reports regressions that add post-COVID × non-core loans to the baseline specifications with dependent variables (1) net worth, (2) delinquent loans, (3) solvency, and (4) net charge-offs. All regressions include the stated controls and fixed effects for year, data system, credit union type, region, and peer group. Standard errors of all specifications are clustered by credit union to account for within-credit-union serial correlation and heteroscedasticity over time. * and *** indicate statistical significance at the 10 percent and 1 percent levels, respectively. Variables are defined in Appendix A.

	(1)	(2)	(3)	(4)
	Net Worth	Delinquent Loans	Solvency	Net Charge-Offs
Post-COVID × Non-Core Loans	0.0129	−0.0133	−0.0939	−0.1439
	(0.054)	(0.123)	(0.079)	(0.096)
Non-Core Loans	−0.1516 ***	0.2835 ***	−0.1356 ***	−0.0973
	(0.034)	(0.089)	(0.050)	(0.064)
Member Size	−0.0169 ***	0.0120 ***	−0.0272 ***	0.0204 ***
	(0.002)	(0.002)	(0.003)	(0.002)
Size	0.0008	−0.0169 ***	0.0003	−0.0123 ***
	(0.002)	(0.003)	(0.003)	(0.002)
Age	0.0143 ***	−0.0127 ***	0.0213 ***	−0.0062 ***
	(0.001)	(0.002)	(0.002)	(0.002)
Land and Building	−0.2010 ***	0.1185 ***	−0.2885 ***	−0.1611 ***
	(0.030)	(0.044)	(0.047)	(0.034)
ROA	0.0789 ***	−0.0826 ***	0.1185 ***	−0.1291 ***
	(0.004)	(0.006)	(0.007)	(0.007)
Operating Expenses	−0.0004	0.1064 ***	0.0119 *	0.0995 ***
	(0.004)	(0.007)	(0.007)	(0.005)
Constant	0.2015 ***	0.2716 ***	1.2873 ***	0.1038 ***
	(0.019)	(0.032)	(0.032)	(0.022)
Year FEs	Yes	Yes	Yes	Yes
Data System FEs	Yes	Yes	Yes	Yes
CU Type FEs	Yes	Yes	Yes	Yes
Region FEs	Yes	Yes	Yes	Yes
Peer Group FEs	Yes	Yes	Yes	Yes
Adjusted R2	0.24	0.13	0.23	0.09
Observations	156,570	156,570	156,570	156,570

. The interaction estimates suggest that the baseline associations are broadly stable across periods. Across all four outcomes, the post-COVID interaction terms are small and statistically insignificant, indicating that the relationships between non-core lending and net worth, delinquent loans, the solvency measure, and net charge-offs do not materially differ in the post-COVID years once common time shocks and persistent differences across credit unions are absorbed by the fixed effects. In particular, we do not find evidence that the association with realized losses becomes systematically stronger after COVID. While non-traditional lending channels may affect how credit risk is realized and recorded, including through loan sales, restructuring, or other loan management practices, the post-COVID interaction results suggest that the overall conditional patterns remain similar across periods in our sample (Keys et al. 2010; Purnanandam 2011).

Regarding growth outcomes, as shown in

Table 9. Post-COVID interaction and growth. This table reports regressions that add post-COVID × non-core loans to the growth specifications with dependent variables (1) loan growth, (2) asset growth, and (3) member growth. All regressions include the stated controls and fixed effects for year, data system, credit union type, region, and peer group. Standard errors of all specifications are clustered by credit union to account for within-credit-union serial correlation and heteroscedasticity over time. ** and *** indicate statistical significance at the 5 percent and 1 percent levels, respectively. Variables are defined in Appendix A.

	(1)	(2)	(3)
	Loan Growth	Asset Growth	Member Growth
Post-COVID × Non-Core Loans	2.8542	2.8906 **	3.2013 ***
	(1.981)	(1.216)	(1.189)
Non-Core Loans	2.8817 ***	0.1778	1.1643 **
	(0.884)	(0.552)	(0.525)
Member Size	0.0109	0.0672 ***	−0.1720 ***
	(0.016)	(0.010)	(0.012)
Size	−0.1035 ***	−0.2244 ***	0.0711 ***
	(0.016)	(0.010)	(0.010)
Age	−0.2375 ***	−0.1244 ***	−0.1629 ***
	(0.021)	(0.012)	(0.014)
Land and Building	3.4939 ***	2.1920 ***	3.3484 ***
	(0.330)	(0.222)	(0.232)
ROA	1.7374 ***	2.5072 ***	1.2406 ***
	(0.063)	(0.040)	(0.035)
Operating Expenses	0.1229 ***	0.2652 ***	0.2879 ***
	(0.043)	(0.027)	(0.027)
Constant	2.7910 ***	3.8986 ***	0.6213 ***
	(0.202)	(0.125)	(0.119)
Year FEs	Yes	Yes	Yes
Data System FEs	Yes	Yes	Yes
CU Type FEs	Yes	Yes	Yes
Region FEs	Yes	Yes	Yes
Peer Group FEs	Yes	Yes	Yes
Adjusted R2	0.10	0.27	0.08
Observations	156,570	156,570	156,570

, the post-COVID interaction term is positive and statistically significant for asset growth and member growth, suggesting that the expansion benefits of non-core activity are stronger in the post-COVID period. Viewed alongside Table 8, this pattern is consistent with the sell and recovery channels discussed above and with the timing of net charge-off recognition since non-core activity can support balance sheet repositioning and expansion while realized losses may be recorded with a lag through workout and charge-off processes. This finding aligns with the broader literature linking intermediaries’ growth strategies to changes in risk taking and portfolio composition, especially around major macroeconomic disruptions (Foos et al. 2010; Berger and Bouwman 2013).

4.8. Discussion

Taken together, the results present a consistent set of patterns linking non-core lending to both growth and financial risk in credit unions. Across the baseline specifications, a higher non-core loan share is associated with faster loan and member growth, indicating that non-core activities are part of how some credit unions expand beyond traditional member lending. At the same time, non-core exposure is linked to weaker buffers, as reflected in a lower net worth and lower solvency, and to weaker loan performance through higher delinquency. The decomposition results sharpen this interpretation by showing that these relationships are not evenly distributed across activities. Loans held for sale appear to be the most strongly connected to both expansion and adverse buffer and credit outcomes, while purchased loans and lease receivables exhibit more limited and less uniform associations.

The interaction estimates also provide an additional layer of interpretation. The adverse associations tied to non-core lending are generally attenuated for larger institutions, whether size is measured by membership or total assets, suggesting that scale can partially moderate the link between non-core exposure and risk outcomes. This does not overturn the main message but it helps reconcile why some large credit unions can operate with meaningful non-core exposure without experiencing the same degree of deterioration observed among their smaller peers.

Overall, the evidence emphasizes that credit union growth strategies cannot be evaluated solely by how quickly loans or membership expand. The composition of lending, and particularly the extent to which growth is driven by non-core components, carries important implications for asset quality, capital strength, and balance sheet resilience, even after controlling for profitability, operating structure, and a comprehensive set of fixed effects.

5. Conclusions

This study investigated how non-core lending is associated with growth, asset quality, and financial resilience in US credit unions using call report data from 1994 to 2024. Non-core lending was measured as the share of purchased loans, lease receivables, and loans held for sale in total loans, and we examined both the aggregate measure and its components. The evidence indicates a clear trade-off. Greater non-core exposure is linked to faster loan growth and member growth, but also to weaker asset quality and thinner buffers, including lower net worth and weaker solvency. The component results show that loans held for sale drive most of these patterns, while purchased loans and lease receivables play smaller and less consistent roles. Post-COVID estimates further suggest that the baseline relationships remain broadly similar, while the growth link is becoming stronger. Our findings imply that lending composition adds information beyond headline growth and is relevant for monitoring credit stress and capital pressure in a sector that builds capital mainly through retained earnings. From a supervisory perspective, the results suggest that risk oversight may benefit from moving beyond growth metrics to incorporate balance sheet composition and the channels through which credit risk is accumulated and absorbed. In particular, elevated non-core shares, especially loans held for sale, can serve as a risk signal that warrants closer attention for underwriting and due diligence, with a focus on smaller institutions where the associations are stronger. Future work can further examine how governance and local market conditions, including field of membership type, geographic environment, and loan product mix, shape non-core strategies and the timing of loss recognition in credit unions.