Borrowing on Belief? Consumer Confidence and U.S. Credit -- A VECM Study

Samiha Tariq; Weikang Zhang

REVIEW 3 major objections 6 minor 12 references

Borrowing on Belief? Consumer Confidence and U.S. Credit -- A VECM Study

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that U.S. consumer confidence and total consumer credit share a stable long-run equilibrium, with confidence persistently driving borrowing while credit adjusts slowly.

desk verdict The paper's own Johansen test and coefficient tables contradict its headline claim of a stable confidence-credit equilibrium, so the VECM results are not usable as reported. read the letter →

arxiv 2505.21832 v1 pith:JD2CMI6B submitted 2025-05-27 econ.GN q-fin.EC

Samiha Tariq , Weikang Zhang This is my paper

classification econ.GNq-fin.EC

keywords consumerconfidencecreditVECMcointegrationimpulseresponsehouseholdborrowingmonetarypolicyerrorcorrection

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Using monthly U.S. data from January 1978 to August 2024, this paper tries to establish that consumer confidence and total consumer credit move together in a stable long-run equilibrium, with confidence as the persistent driver of borrowing. It fits a six-variable vector error-correction model—credit, sentiment, interest rates, inflation, money supply, and unemployment—and reports two cointegrating relationships. The long-run relation is positive: higher confidence is associated with more credit utilization, and an unexpected jump in confidence produces a sustained rise in credit in the impulse-response analysis. In the short run, credit is inertial and responds mainly to money-supply movements rather than to sentiment, and the error-correction speed for credit is about 0.1 percent per month. If correct, the paper implies that household optimism is not just a mirror of the economy but an active, slow-moving input into the credit cycle, with consequences for how monetary and communication policies transmit to household balance sheets.

What carries the argument

The central object is a six-variable Vector Error Correction Model (VECM)—a system that separates short-run dynamics from long-run equilibrium by writing first differences as a function of lagged levels through a matrix $\Pi=\alpha\beta'$. The Johansen trace test selects the cointegration rank, the $\beta$ vectors are the long-run equilibrium relationships, the $\alpha$ loadings are the error-correction speeds, and impulse-response functions trace the dynamic feedback. The VECM is the vehicle that converts 'confidence and credit move together' into a testable claim with a specific, slow speed of return to equilibrium.

What would settle it

Re-estimate the same six-variable system with the Johansen trace test and check whether the restriction $r=2$ is rejected against $r=5$ or full rank; if it is, a levels VAR or a different normalization should replace the error-correction model. Alternatively, drop confidence from the system and test whether the remaining five variables still produce the same credit equilibrium; if they do, confidence is not load-bearing.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is a stable, two-dimensional long-run equilibrium between U.S. consumer credit and consumer confidence, controlling for the federal funds rate, CPI inflation, M2, and unemployment. The first normalized cointegrating vector is interpreted as a credit equation in which inflation strongly raises credit and monetary conditions restrain it; the second normalizes on consumer confidence, tying it positively to credit and inflation and negatively to the funds rate. Adjustment is slow: the credit equation's error-correction coefficient is $-0.001$ per month, and confidence corrects even more slowly, so deviations from equilibrium fade over many months. Impulse responses show that a one-standard-deviation shock to confidence raises credit persistently, while a credit shock initially lowers confidence and then recedes, which the paper reads as a stable long-run complementarity with asymmetric feedback.

Load-bearing premise

The load-bearing premise is that the six series are non-stationary in levels but stationary in differences and that exactly two cointegrating relationships exist; if the true rank is different—the paper's own trace test rejects even the hypothesis of four or fewer—the two long-run equations and the impulse responses are not valid equilibrium objects.

Editorial extensions

If this is right

In the long run, a rise in consumer confidence should raise total consumer credit persistently, so sentiment becomes a candidate channel for credit-cycle policy.
Short-run credit management operates chiefly through liquidity conditions such as M2, not through sentiment, which matters for the timing of policy effects.
Because credit's error-correction speed is about 0.1 percent per month, shocks to the equilibrium relationship have long half-lives and policy corrections will be slow.
A sudden expansion of credit lowers confidence only temporarily, so debt accumulation is not modeled as a durable psychological drag.
If confidence and credit are truly cointegrated, then policies or communications that shift confidence have multi-year leverage over household borrowing.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

An alternative reading of the paper's own trace test is that the system may be stationary in levels, since every null up to $r\le 4$ is rejected; a levels-VAR comparison would tell whether the imposed rank of two is the right lens.
The credit-normalized cointegrating vector in the appendix carries a near-zero coefficient on confidence, so the 'confidence raises credit' interpretation rests on the second vector; testing the restriction that confidence enters the first vector would sharpen the claim.
A testable extension is to split the sample around major regimes such as the 2008 financial crisis and the COVID-19 pandemic and ask whether the cointegrating slope and error-correction speed are stable; if they shift, the single full-sample equilibrium may be a composite of regimes.
Because M2 enters with opposite signs at lags one and two in the credit equation, the implied liquidity effect is oscillatory; a structural shock decomposition on M2 would clarify whether the net effect is expansionary.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

This one is a good example of a paper whose own appendix contradicts its abstract. The idea is reasonable—take FRED data from 1978 to 2024, put consumer credit and confidence in a VECM with macro controls, and look for long-run links. The writing is clear, the literature review is appropriate (Lamdin 2008 is the relevant benchmark), and a long-sample VECM with additional controls could be a useful incremental contribution.

The problem is load-bearing. The Johansen trace test in Table 3 rejects every null from r≤0 through r≤4—including full rank. For a five-variable system that means the levels are stationary, so the I(1) cointegration framework itself is on shaky ground. The paper openly acknowledges this and then says it imposes r=2 by 'balancing statistical evidence with economic interpretability.' That is not a rank-selection rule; it is picking the answer that fits the story. With an indefensible rank, the cointegrating vectors and IRFs are not identified as equilibrium objects.

And the reported numbers don't support the prose anyway. Table 10 gives consumer confidence a coefficient of 1.890e-13 with z=0.000; Table 11 gives credit a coefficient of 4.780e-20. Those are zeros. The long-run positive confidence-credit link described in the abstract is not present in the paper's own estimates. There are also data description inconsistencies: UNRATE appears in the methodology and in Table 2, but not in the VECM equations (Tables 4–8); Table 2 treats FEDFUNDS and UNRATE as stationary in levels, which is odd for a VECM; and inflation is described as 'monthly percentage changes' yet the tables use the CPI level. The sample sizes also differ across series (788 vs 634).

What's worth saving: the question is real, the data are public, and a careful redo with a defensible rank selection procedure could be a decent empirical paper. But as submitted, the central claims are contradicted by the paper's own output. This is not a matter of taste; it's a specification failure. I wouldn't bring it to reading group, wouldn't cite it, and if I were the editor I'd desk reject rather than spend referee time. A revision would need to start with the rank test.

Referee Report

3 major / 6 minor

Summary. The paper estimates a five-variable VECM linking U.S. total consumer credit, the University of Michigan consumer sentiment index, the federal funds rate, CPI inflation, and M2 money supply, using monthly FRED data claimed to span January 1978 to August 2024. It reports Johansen trace tests, two normalized cointegrating vectors, short-run coefficient tables, loading coefficients, and impulse-response functions, and concludes that there is a stable long-run equilibrium in which higher consumer confidence raises credit utilization. The paper also claims slow error correction and persistent positive impulse responses from confidence to credit.

Significance. If the central claim were established, the paper would contribute a useful long-sample, macro-controlled description of how sentiment and household credit interact. The choice of a long monthly sample and the inclusion of monetary and price controls are sensible, and the paper explicitly reports its unit-root and cointegration pretests. However, the manuscript as written does not establish the claimed result: the Johansen test rejects full rank for the five-variable system, the imposed rank of two is not justified by any valid selection rule, and the reported cointegrating vectors contain effectively zero confidence and credit coefficients. These problems are load-bearing rather than cosmetic, so the paper's headline conclusions are not supported by its own evidence. No replication code or data are provided, and several appendix tables contain internal inconsistencies.

major comments (3)

[§4.1, Table 3] The Johansen trace test in Table 3 rejects every null hypothesis from r≤0 through r≤4, with trace statistics 143.462, 72.958, 30.582, 17.220, and 5.777 all exceeding the 5% critical values. For the five-variable system estimated in Tables 4–8, rejecting r≤4 implies full rank, i.e., the levels are stationary, which contradicts the I(1) assumption required for a VECM. The paper's stated response—imposing r=2 by 'balancing statistical evidence with economic interpretability'—is not a statistical rank-selection procedure. Because the rank determines whether the cointegrating vectors and adjustment coefficients are identified as long-run equilibrium parameters, the reported β vectors and the impulse-response functions in §4.4 are not valid as presented. The authors must reconcile the ADF pretests with the cointegration test or re-estimate under a defensible rank selection rule.
[Tables 9–11] The reported cointegrating vectors do not contain the positive confidence–credit relationship claimed in the abstract and in §4.2.1. Table 10 reports a Consumer Confidence coefficient of 1.890e-13 with z=0.000 in the first cointegrating vector, and Table 11 reports a credit coefficient of 4.780e-20 in the second vector; as printed, these are zeros. The narrative in §4.2.1 describes the first vector as showing that higher consumer confidence raises total consumer credit, but the table does not support that reading. In addition, Table 9 reports an error-correction loading of 137.156 for total credit on ec2, which is inconsistent with the 'slow adjustment' description in §4.2.3. These tables need to be corrected, and the text must be rewritten to match the actual estimates.
[§3, Eq. (1), Tables 1 and 4–8] The data description is internally inconsistent. Section 3 and Eq. (1) state that the unemployment rate (UNRATE) is one of the endogenous variables, but no UNRATE equation appears in Tables 4–8 and Table 9 reports loading coefficients for only five equations. Table 1 reports 788 observations for most variables and 634 for consumer confidence, whereas the stated monthly sample January 1978–August 2024 contains 560 observations. The authors must correct the sample description and either include UNRATE in the estimated system or explicitly state that it was excluded.

minor comments (6)

[§4.4, Figure 2] Figure 2 is referenced in §4.4, but no figure appears in the manuscript, so the reader cannot verify the impulse-response paths or the 95% confidence bands.
[Table 3] The paper does not state which deterministic specification was used for the Johansen trace test (for example, whether a constant was restricted to the cointegrating space); the critical values depend on this choice and should be reported.
[Table 1] Table 1 labels monetary figures as being in millions of dollars, while the introduction and abstract use trillions; the units should be made consistent across the paper.
[Table 11] Table 11 uses parameter labels β1–β5 without the corresponding variable names, unlike Table 10, making the two cointegrating vectors difficult to compare; variable labels should be added.
[Section 2] The paper cites Kahneman and Tversky (2013) for Prospect Theory, but the canonical reference is Kahneman and Tversky (1979); please correct the citation.
[Tables 10–11] Several coefficients are reported with standard errors and p-values of exactly 0.000, and some coefficients are given as e-13 or e-20 with identical lower and upper confidence bounds; these should be reported to a meaningful precision or flagged as estimation issues.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: all estimates come from external FRED data; the rank restriction is a statistical assumption, not a circular step.

full rationale

The paper's central quantities—cointegrating vectors, error-correction terms, and impulse responses—are all computed from a VECM estimated on monthly FRED time series from January 1978 through August 2024. None of these quantities is defined in terms of the paper's target claim, and no parameter is fit to a subset of data and then 'predicted' on a closely related quantity. The Johansen trace test rejects r≤4, and the authors impose r=2 'balancing statistical evidence with economic interpretability'; that is a model-selection assumption, and if the rank is wrong the cointegrating vectors and impulse responses are unidentified, but this is a statistical-validity issue, not circular reasoning. The statement that impulse-response analysis 'confirms' the long-run relationship is a description of in-sample implications of the fitted model rather than an independent confirmation; however, reporting IRFs derived from an estimated model is standard empirical practice and does not reduce the derivation to an equation tautology. There are no load-bearing self-citations and no imported uniqueness arguments. Therefore no significant circularity is present.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central inference rests on a two-rank VECM chosen after the trace test rejected every rank including full rank, plus stable-parameter assumptions across 1978-2024. The free parameters are the imposed rank and lag order; the axioms are the standard time-series assumptions plus the unexamined stability assumption.

free parameters (2)

Cointegration rank r = 2 (imposed)
Johansen trace test rejects r<=0 through r<=4, i.e. suggests full rank; authors impose r=2 for economic interpretability in Section 4.1.
Lag order p = 2
Selected by SIC and HQIC in Section 4.1; standard but data-dependent.

assumptions (4)

domain assumption The six series are I(1) and jointly follow a linear cointegrating VECM with constant parameters.
ADF tests in Table 2 classify credit, sentiment, CPI, and M2 as non-stationary and FEDFUNDS and UNRATE as stationary, but the trace test rejects all ranks, inconsistent with the I(1) setup.
standard math Johansen trace test asymptotic critical values apply to this sample.
Used to determine cointegration; no small-sample or structural-break corrections are reported.
domain assumption UMCSENT is a valid proxy for consumer confidence.
The University of Michigan index is used without discussion of measurement error or alternative sentiment measures.
domain assumption No structural breaks, such as the 2008 crisis or COVID-19, alter the long-run parameters.
The sample spans both crises but no break tests or dummy variables are reported, yet long-run stability is claimed.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Borrowing on Belief? Consumer Confidence and U.S. Credit -- A VECM Study." pith.science (2026). https://pith.science/paper/JD2CMI6B

@misc{pith2026250521832,
  author       = {Pith},
  title        = {Pith review of: Borrowing on Belief? Consumer Confidence and U.S. Credit -- A VECM Study},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JD2CMI6B}},
  note         = {Machine review of arXiv:2505.21832}
}

read the original abstract

This study explores the interdependent relationship between consumer credit and consumer confidence in the United States using monthly data from January 1978 to August 2024. Utilizing a Vector Error Correction Model (VECM), the analysis focuses on the interplay between household borrowing behaviour and consumer sentiment while controlling for macroeconomic factors such as interest rates, inflation, unemployment, and money supply. The results reveal a stable long-run equilibrium: heightened consumer confidence is associated with increased credit utilization, reflecting greater financial optimism among households. In the short run, shifts in consumer confidence exert relatively modest immediate influence on credit usage, whereas consumer credit adjusts slowly, displaying significant inertia. Impulse-response analysis confirms that shocks to consumer confidence generate sustained positive effects on borrowing, while unexpected increases in credit initially depress sentiment but only fleetingly. These findings underscore the critical role of the relationship between consumer confidence and credit-market dynamics and highlight its policy relevance for fostering balanced and stable household finances.

Figures

Figures reproduced from arXiv: 2505.21832 by the authors.

**Figure 2.** Impulse–response functions from the VECM, [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

12 extracted references · 11 canonical work pages

[1]

Borrowing on Belief ? Consumer Confidence and U.S. Credit—A VECM Study Samiha Tariq Weikang Zhang Southern Illinois University Carbondale May 2025 Abstract This study explores the interdependent relationship be- tween consumer credit and consumer confidence in the United States using monthly data from January 1978 to August

work page 2025
[2]

Figure 1: Time–series evolution of key U.S

tests confirm that credit, consumer sentiment, CPI, and M2 are non-stationary, while the federal-funds and unemployment rates are stationary af- ter first differencing. Figure 1: Time–series evolution of key U.S. macro- financial indicators, 1960–2024: total consumer credit, consumer confidence, federal funds rate, CPI, money sup- ply (M2), and unemployme...

work page 1960
[4]

This comprehensive dataset captures multiple economic cycles and significant events, including the 2008 financial crisis and the COVID-19 pan- demic, enabling a robust analysis of consumer-credit be- haviour trends. Unlike previous studies constrained by fewer variables or shorter timeframes, this research in- corporates consumer confidence, interest rate...

work page 2013
[5]

is more sensitive to contemporaneous macro factors. Significant drivers include its own first lag (−0.1034,p <0.05), the second lag of the federal funds rate (−0.8939,p <0.01), the first lag of CPI (−1.0467,p <0.01), and the first (−0.0095,p <0.05) and second (0.0063,p <0.10) lags of M2. 4.2.3 Error-Correction Mechanism The adjustment coefficients confirm...

work page 1978
[7]

The empir- ical results indicate the presence of two stable long-run relationships, underscoring the deep structural linkage be- tween household borrowing behavior and consumer sen- timent. Specifically, higher levels of consumer confidence significantly encourage greater credit utilization in the long run, while tighter monetary conditions and rising inf...

arXiv 2023
[11]

(2025).US total consumer credit outstand- ing

YCharts. (2025).US total consumer credit outstand- ing. https://ycharts.com/indicators/total_consumer_ credit_outstanding 6 Appendix Table 1: Descriptive Statistics Variable Observ. Mean Std. Dev. Min Max Total consumer credit 788.000 1 454 936.979 1 452 142.626 48 961.160 5 097 589.770 Consumer Confidence 634.000 85.217 13.037 50.000 112.000 Interest Rat...

work page 2025
[12]

Null Hypothesis Trace statistic 95% critical value r≤0 143.462 69.819 r≤1 72.958 47.855 r≤2 30.582 29.796 r≤3 17.220 15.494 r≤4 5.777 3.841 Table 4: Lagged Parameters for Equation: Total Consumer Credit (Level) Regressor Coef. Std. Err.z P >|z|95% CI Lower 95% CI Upper L1.Total consumer credit 0.127 0.042 3.054 0.002 0.046 0.209 L1.UMCSENT−3.586 109.010−0...

work page
[257]

https://doi.org/10.1080/13504851.2017.1321836 Federal Reserve Bank of New York. (2020). textitQuar- terly report on household debt and credit: 2019 Q4. Federal Reserve Bank of New York. (2024).Household debt and credit report: 2024 Q4. Federal Reserve. (2020, March 15).FOMC statement. https://www.federalreserve.gov/monetarypolicy/fomcminutes20200315.htm F...

work page arXiv 2020

Show all 12 references

[2006]

The findings from regression and Granger-causality analyses confirm a unidirectional causal relationship from con- sumer confidence to household credit-card expenditures

The study assesses how consumer confidence, indicative of future financial and economic expectations, affects spending behaviours. The findings from regression and Granger-causality analyses confirm a unidirectional causal relationship from con- sumer confidence to household c...

2022
[2020]

(2025, May 3)

https://www.bls.gov/opub/ted/2020/unemployment- rate-rises-to-record-high-14-point-7-percent-in-april- 2020.htm Bureau of Labor Statistics. (2025, May 3). The employment situation—April

2020
[2024]

Utilizing a Vector Error Correction Model (VECM), the analysis focuses on the interplay between household borrowing behaviour and consumer sentiment whilecontrollingfor macroeconomic factors such as in- terest rates, inflation, unemployment, and money sup- ply. The results rev...

2023 arXiv
[2025]

https://www.bls.gov/news.release/pdf/empsit.pdf Dow, J. P. (2018). Attitudes towards credit after the Great Recession.Applied Economics Letters, 25(4), 254–

2018

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Borrowing on Belief ? Consumer Confidence and U.S. Credit—A VECM Study Samiha Tariq Weikang Zhang Southern Illinois University Carbondale May 2025 Abstract This study explores the interdependent relationship be- tween consumer credit and consumer confidence in the United States using monthly data from January 1978 to August

work page 2025

[2] [2]

Figure 1: Time–series evolution of key U.S

tests confirm that credit, consumer sentiment, CPI, and M2 are non-stationary, while the federal-funds and unemployment rates are stationary af- ter first differencing. Figure 1: Time–series evolution of key U.S. macro- financial indicators, 1960–2024: total consumer credit, consumer confidence, federal funds rate, CPI, money sup- ply (M2), and unemployme...

work page 1960

[3] [4]

This comprehensive dataset captures multiple economic cycles and significant events, including the 2008 financial crisis and the COVID-19 pan- demic, enabling a robust analysis of consumer-credit be- haviour trends. Unlike previous studies constrained by fewer variables or shorter timeframes, this research in- corporates consumer confidence, interest rate...

work page 2013

[4] [5]

is more sensitive to contemporaneous macro factors. Significant drivers include its own first lag (−0.1034,p <0.05), the second lag of the federal funds rate (−0.8939,p <0.01), the first lag of CPI (−1.0467,p <0.01), and the first (−0.0095,p <0.05) and second (0.0063,p <0.10) lags of M2. 4.2.3 Error-Correction Mechanism The adjustment coefficients confirm...

work page 1978

[5] [7]

The empir- ical results indicate the presence of two stable long-run relationships, underscoring the deep structural linkage be- tween household borrowing behavior and consumer sen- timent. Specifically, higher levels of consumer confidence significantly encourage greater credit utilization in the long run, while tighter monetary conditions and rising inf...

arXiv 2023

[6] [11]

(2025).US total consumer credit outstand- ing

YCharts. (2025).US total consumer credit outstand- ing. https://ycharts.com/indicators/total_consumer_ credit_outstanding 6 Appendix Table 1: Descriptive Statistics Variable Observ. Mean Std. Dev. Min Max Total consumer credit 788.000 1 454 936.979 1 452 142.626 48 961.160 5 097 589.770 Consumer Confidence 634.000 85.217 13.037 50.000 112.000 Interest Rat...

work page 2025

[7] [12]

Null Hypothesis Trace statistic 95% critical value r≤0 143.462 69.819 r≤1 72.958 47.855 r≤2 30.582 29.796 r≤3 17.220 15.494 r≤4 5.777 3.841 Table 4: Lagged Parameters for Equation: Total Consumer Credit (Level) Regressor Coef. Std. Err.z P >|z|95% CI Lower 95% CI Upper L1.Total consumer credit 0.127 0.042 3.054 0.002 0.046 0.209 L1.UMCSENT−3.586 109.010−0...

work page

[8] [257]

https://doi.org/10.1080/13504851.2017.1321836 Federal Reserve Bank of New York. (2020). textitQuar- terly report on household debt and credit: 2019 Q4. Federal Reserve Bank of New York. (2024).Household debt and credit report: 2024 Q4. Federal Reserve. (2020, March 15).FOMC statement. https://www.federalreserve.gov/monetarypolicy/fomcminutes20200315.htm F...

work page arXiv 2020