INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 218

Modeling the Impact of Macroeconomic Variables on Personal

Loan Applications and Approvals Before and During the COVID-

19 Pandemic
Nur Arina Bazilah Kamisan1, Lee Yin Jun2, Siti Mariam Norrulashikin3 and Atikah Abu4

1 Faculty of Computing, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor, Malaysia.

2 Department of Mathematical Sciences, Faculty of Science, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor,

Malaysia.

3 Department of Mathematical Sciences, Faculty of Science, Universiti Teknologi Malaysia, 81310 Johor Bahru, Johor,

Malaysia.

4 Centre for Mathematical Sciences, Universiti Malaysia Pahang Al-Sultan Abdullah, Lebuh Persiaran Tun Khalil Yaakob,

26300 Kuantan, Pahang, Malaysia.

DOI: https://doi.org/10.51583/IJLTEMAS.2025.1409000030

Received: 29 Aug 2025; Accepted: 05 Sep 2025; Published: 01 October

Abstract: This study investigates the association between personal loan applications and approvals and macroeconomic variables,
including bank interest rates, the consumer price index (CPI), and employment and unemployment figures, both prior to and during
the COVID-19 pandemic. Data spanning January 2018 to June 2022, were utilized for the analysis. Two analytical techniques were
employed: the correlation matrix and the multiple linear regression model. The correlation analysis revealed strong relationships
between the CPI and employment figures, as well as between interest rates and unemployment, in their association with both loan
applications and approvals. Results from the regression model indicated that bank interest rates and employment status significantly
influenced personal loan applications, regardless of the pandemic context. With respect to loan approvals, the findings demonstrated
that bank interest rates, unemployment, and the CPI exerted significant effects, particularly in regions severely affected by COVID-
19. These results highlight the critical role of macroeconomic conditions in shaping lending behavior and provide insights into the
interaction between financial markets and economic shocks.

Keywords: Personal loan; consumer price index; correlation; COVID-19; pandemic; multiple linear regression; employment rate;
bank interest.

I. Introduction

Providers of financial goods have observed the impact of the Covid-19 outbreak on consumer behaviour. During this time, there has been
a significant shift toward online operations due to restrictions on freedom of movement and the risk to consumer health. Interpersonal trust
and social media use have a positive and significant impact on e-commerce application usage. Market size, consumer capacity, and
commercial infrastructure all demonstrated positive effects, albeit to variable degrees. However, market intensity did not have a statistically
meaningful impact. The study by Valdivino et al. (2025) suggests that the use of e-commerce applications in Latin America is due to a
combination of individual and systemic factors. Access to e-commerce applications is governed by commercial infrastructure and
economic competence, but consumer adoption is directly influenced by interpersonal trust and social media activity.

In Malaysia, the ratio of household debt to disposable income stands at 140.4%, significantly higher than 105.3% in Singapore, 123.3%
in the United States, and 52.7% in Thailand, making Malaysia one of the highest globally in this metric. This indicates that Malaysian
households, on average, borrow an amount equivalent to 1.4 times their annual income (Ismail et al., 2018). The early months of 2020
saw food shortages driving inflation, while disruptions in production and consumption across various industries created temporary

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 219

bottlenecks in the supply chain, leading to significant innovations in supply chain management. The need for pandemic containment
further compelled individuals to save and seek alternative income sources (Oindrila Chakraborty, 2022).

The "buy now, pay later" trend has led to increased credit card usage among Malaysian Generation Y (Gen Y). Research shows that
knowledge of credit cards and self-efficacy are negatively associated with credit card misuse among Gen Y in Malaysia, while attitudes
toward credit cards, materialism, and social norms are positively correlated with misuse (Zainudin et al., 2019). To address these challenges,
Uriawan et al. (2019) proposed a collateral-light peer-to-peer personal lending platform. The trustworthiness of borrowers emerges as a
critical predictor of on-time payments, with reputation and interpersonal relationships serving as key factors in assessing borrower
reliability.

The COVID-19 pandemic has significantly hindered banks' capacity to meet credit demand and provide liquidity and funding. Global
lockdowns and border restrictions have aimed to reduce health-related expenses and ensure public safety (Najaf et al., 2022). Kamarudin
et al. (2021) explores the pricing of goods and services across various sub-sectors, with a particular focus on essential items needed by all
consumers. Throughout the pandemic, prices generally declined, although the consumer price index (CPI) for food increased, reflecting
similar trends in health and education sectors (Kamarudin et al., 2021). Sales of consumer durable goods rose during the pandemic, driven
by consumer preferences and disposable incomes. However, these trends are expected to diminish as public health concerns are addressed,
and the U.S. economy reopens (Tauber & Van Zandweghe, 2021). Gur (2024) implied Long Short-Term Memory (LSTM), MARS,
XGBoost, and hybrid approaches like LSTM-MARS and LSTM-XGBoost, to enhance the forecasting accuracy of the US CPI and found
that variables such as past values of the CPI, oil prices, and gross domestic product (GDP) significantly influence CPI.

The pandemic has also led to increased loan delinquency, attributed to high interest rates and business failures. Consequently, banks may
find their relevance challenged, with their credit culture and lending variables becoming more critical in the context of priority sector
lending (Kumar Tiwari & Bapat, 2020). Interestingly, rising oil prices and new COVID-19 cases have positively influenced the Saudi
banking index, despite the impacts of lockdowns and interest rate cuts (Assous & Al-Najjar, 2021). Study from Narvekar & Guha (2021),
enhanced bankruptcy prediction research by employing machine learning techniques, particularly XGBoost, on a quarterly dataset of
financial ratios from U.S. public firms (1970 – 2019) and applies this model to forecast a substantial rise in bankruptcy rates in the second
half of 2020 due to COVID-19 lockdowns, suggesting that the levels may not significantly exceed those of 2010.

A large data set on COVID-19 relief policy loans in Taiwan was used to examine the correlations between the speed of loan approval for
individual applicants and their demographic characteristics, credit data, banking relationships, and the corporate social responsibility of
the lending banks by Kung et al. in 2024. The influence of lending bank characteristics on the financial inclusion of applicants with poor
credit was also analyzed. Evidence was found that the prioritization of individuals in need enabled the policy to be implemented in a
manner that facilitated the rapid disbursement of loans, thereby substantially promoting financial inclusion (Kung et al., 2024).

The primary objectives of this research are to compare the effects of bank interest rates, employment status, and the consumer price index
on the likelihood of approval for personal loans from Bank Negara Malaysia, both before and during the COVID-19 pandemic. This will
be achieved using correlation methods and multiple linear regression models.

II. Methodology

Multiple regression analysis can be looked upon as an extension of straight-line regression analysis which involves only one
independent variable, to the situation in which more than one independent variable must be considered (David G. Kleinbaum, 2014).
The general multiple regression model can be represented by the following equation:

�� = ��0 + ��₁��₁ + ��₂��₂+. . . . . . + �� + �� (1)

where

��₁ �� are independent variables

��0, ��1, ��2 , . . . . . . , �� are regression coefficients

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 220

�� is the predicted value of the dependent variable

�� is the is the model’s random error (residual) term (how much variation there is in predicted ��).

The model takes from �� = ��ᵀ�� + ��. Where ᵀ denotes the transpose, so that ��ᵀ�� is the inner product between vectors
��ᵀ and ��. Often these �� equations are stacked together and written in matrix notation as:

�� = �� + ��. (2)

where

�� = [

��1
��2
⋮
��

]

�� =

[

��1ᵀ
��2ᵀ
⋮

��ᵀ]

=

[

1 ��11 ⋯ ��1��
1 ��21 ⋯ ��2��
⋮ ⋮ ⋱ ⋮
1 ��1 ⋯ ��]

�� =

[

��0
��1
��2
⋮

��]

�� = [

��1
��2
⋮
��

The p-values help determine whether the associations observed in your sample are present in the larger population. To apply
regression analysis, a model must be fitted and validated. Examine then the regression coefficients and ��-values using. Low ��-
values indicate statistical significance for the independent variable (typically 0.05). The coefficients represent the average change
in the dependent variable when the independent variable (IV) is altered by one unit, while the other IVs remain unchanged (Forst,
2017).

2.1. Correlation Coefficient

Correlation coefficient provides a measure of how two random variables are linearly associated in a sample and has properties
closely related to those of straight-line regression. Sample correlation coefficient,

�� =
��(∑��) − (∑��)(∑��)

√(��∑��2 − (∑��)2)(��∑��2 − (∑��)2
. (3)

Least-square estimate of the slope of fitted regression line,

�� =
��
��

��̂. (4)

The relationship becomes more positive as r becomes more positive. This implies that, when r is close to 1, a person with a high
value for one variable will likely have a high value for the other variable, and a person with a low value for one variable will likely
have a low value for the other variable. The squared correlation coefficient or coefficient of determination ��2 measures the strength
of the linear relationship between the dependent variable Y and the independent variable X. The closer ��2 is to 1, the stronger the
linear relationship; the closer ��2 is to 0, the weaker the linear relationship.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 221

2.2. ANOVA

To summarize the results of a multiple regression analysis, the analysis of variance (ANOVA) technique is often employed. The
specific form of an ANOVA table may differ, however, depending on how the contributions of the independent variables are to be
regarded (individually or collectively).

Coefficient of determination, ��2 =
�� − ��

��
. (5)

Regression sums of squares,

(�� − ��) = ∑(��̂�� − ��̅)
2

∞

��=1

. (6)

Total unexplained variation = Variation due to regression + Residual variation after regression.

∑(�� − �� )²
∞

��=1

= ∑(��̂�� − �� )² +
∞

��=1

∑( �� − ��̂�� )²
∞

��=1

(7)

Sum of squares = Regression sums of squares + Residual sum of squares. The test is used to check if there is a linear relationship
between the dependence variable and at least one of the predictor variables. The hypotheses are shown below:

��0 = ��0 = ��1 = ⋯ = �� = 0

��1 = �� ≠ 0 ��

The ��0 will be rejected if the calculated F value is bigger than ��,��,(��−(��+1)). Rejection of ��0 implies that the regression

coefficient differs from zero. That is at least one predictor variable is significant. The significance �� value showed in the ANOVA
table can also compare with the significance level, �� value. If the significance value F is smaller than ��, then the null hypothesis
can be rejected. The way of testing ��0 is equivalent to using the two-sided t test. This is so because, for �� degrees of freedom,

��1,�� = ��
2 (8)

��1,��,1−�� = ��,1−��/2
2 (9)

The expression states that the 100(1 − ��)% point of the F distribution with 1 and �� degree of freedom is the same as the square
of the 100(1 − ��/2)% point of the t distribution with �� degree of freedom. The calculation of the ANOVA is explained in Table
1 below.

Table 1. The calculation of every value generated in ANOVA table.

Source Degree of freedom, DF Sum of Squares, SS Mean Square, MS F-value

Model (Regression) �� − �� Regression
�� Regression

�� Regression
�� Residual

Error (Residual) �� − �� − 1 ��
�� Residual

Corrected Total �� − 1 ��

where �� is the number of parameters, �� is the number of data, �� is the total sum of squares, and �� is the sum of squares

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 222

due to error.

2.3. t-test

When employing a multiple linear regression model to ascertain whether a linear relationship exists between the response and
predictor variables, the t-test is utilized to determine the statistical significance of the various regression coefficients. The hypothesis
statements to test the significance of a particular regression coefficient, ��

��0 = �� = 0

��1 = �� ≠ 0.

The test statistics for this test are based on t distribution:

�� =
��̂��

��̂��

(10)

where ��̂�� is the corresponding estimated coefficient and ��̂��
is the estimate of the standard error, both are produced by standard

regression programs. ��0 will be rejected if the t-test lies outside the acceptance region:

 |��| > ��−��−2,1−
��
2
(two-sided test; ��: �� ≠ 0) Test for linearity

 �� > ��−��−2,1−�� (right-tailed test: ��: �� > 0) Test for a positive slope

 �� > ��−��−2,1−�� (left-tailed test; ��: �� < 0) Test for a negative slope

Rejection ��0 implies that the �� is significant to the model. Besides, we can also draw the conclusion from p-value. If the p-

value is smaller than the significance level, �� = 0.05, the null hypothesis is rejected. Hence, we have sufficient evidence to
conclude that the variable is significantly contributing to the model.

III. Results and Discussion

In this study, the data is obtained from Bank Negara Malaysia (BNM) website except for COVID-19 cases, where the data is
obtained from Kementerian Kesihatan Malaysia (KKM) website. The personal loan data consists of the number of applications and
number of approvals from BNM as the dependent variable and 5 independent variables which are interest rate, employed number,
unemployed number, CPI and COVID-19 cases. The definition of the labels used in this study is listed in Table 2.

Table 2. The response (dependent variables) and explanatory variables (independent variables) used in this study.

Labels Definitions

Response Variable

�� Personal loan approvals

�� Personal loan applications

Continuous Explanatory Variables

Interest BNM personal loan interest

Employed Monthly record: Malaysians get employment

Unemployed Monthly record: Malaysian loss of job or retirement

CPI Monthly record: Consumer Price Index

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 223

COVID-19 Monthly new confirmed cases of COVID-19 in Malaysia

The personal loan applied and approved will be studied to see the relation between these two variables with interest rate, employed
and unemployed number, CPI and COVID-19 cases. The monthly data for all these variables is recorded monthly. The plots for
loan applied and approved can be shown below.

Figure 2. Time series plot of personal loan applied, and personal loan approved.

The graph in Figure 1 above displays the number of banks applied and approved personal loans in Malaysia. It is apparent that the
loan applied reduces during the early stages of COVID-19, but it returns to normal after July 2020. The loan applied has a larger
decrement than the loan granted. This is understandable given that the quantity of applications has been reduced throughout this
period. Loans applied vary more than loans approved. It is it is apparent banks have targeted figures when issuing loans every
month. The descriptive statistics of the response variables which are the personal loan applied and approved is shown in Table 3.

Table 3. Descriptive statistics of the response variables; personal loan application and personal loan approval.

Loan Application Loan Approval

Mean 5330.7671 1965.8020

Standard Error 131.2442 51.2939

Median 5467.5455 1985.6375

Mode - -

Standard Deviation 964.4442 376.9317

Sample Variance 930152.5655 142077.5060

Kurtosis 4.5269 2.3005

Skewness 1.6369 -1.1509

Range 5044.6490 1868.4230

Maximum 6701.1560 2528.0170

Minimum 1656.5070 659.5940

Sum 287861.4210 106153.3080

Count 54 54

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 224

Based on descriptive statistics in Table 3, the mean value shows that the loan approved is about 37% from the loan applied. The
standard deviation values show that the loan applied has larger value compared to loan approved which means that the loan
application is more divergent compared to loan approved. It can be supported by the minimum value and the maximum value of
these two variables. The difference between the minimum and maximum number for loan applied is larger than loan approved. As
for the other variables, the descriptive statistics can be shown in Table 4 below.

Table 4. Descriptive statistics of the explanatory variables; unemployment rate, employment rate, bank interest, CPI and COVID-
19 cases.

This table presents descriptive statistics for Unemployment, Employment, Bank Interest, CPI, and COVID-19 cases.
Unemployment has a mean of 625.29 and moderate variability, while Employment is much higher, averaging 15,213 with a wider
range, indicating more fluctuation in the employed population. Bank Interest rates are stable, with a low standard deviation around
a 6.17% average. CPI shows modest variation with a mean of 121.85, suggesting relative price stability. COVID-19 has a high
mean; 136,323 and extreme variability, shown by a large standard deviation; 190,214.27 and range; 762,952, reflecting the impact
of the pandemic. The skewness and kurtosis values indicate that COVID-19 data is positively skewed and has heavy tails, whereas
Bank Interest and Unemployment data are more symmetric and flatter. To observe the relationship between all variables, the
correlation matrix will be utilized. The correlation between every conceivable pair of variables is shown by the correlation matrix.
It is an effective tool for finding and visualizing trends in the data provided as well as for summarizing huge datasets. The correlation
matrix for the personal loan applied and approved is shown in Figure 2 below.

Unemployment Employment Bank Interest CPI COVID-19

Mean 625.2859 15213.0309 6.1716 121.8463 136323.162

Standard Error 15.3248 41.6598 0.0873 0.2728 31271.0331

Median 620.5685 15201.6000 6.2613 121.4000 62085

Standard Deviation 112.6140 306.1358 0.6416 2.0049 190214.268

Sample Variance 12681.9186 93719.1276 0.4117 4.0195 3.62E+10

Kurtosis -1.7193 0.0752 -1.9334 0.7270 3.6430

Skewness 0.2334 0.5558 0.0112 0.6825 2.0554

Range 321.3000 1265.8000 1.4325 9.8000 762952

Maximum 826.1000 14670.5000 6.9221 127.4000 763289

Minimum 504.8000 15936.3000 5.4896 117.6000 337

Sum 33765.4380 821503.6680 333.2669 6579.7000 5043957

Count 54 54 54 54 37

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 225

(a)

(b)

Figure 2. The correlation matrix for (a) personal loan applied and (b) personal loan approved.

Figure 2 above is a correlation matrix visualized with a heatmap and correlation coefficients, showing the relationships between
personal loan applied and different economic and social factors. CPI and employment number have a correlation value of 0.87
which indicates a strong positive correlation between CPI and employment. Interest and unemployment have a correlation value of
-0.92 and this is a very strong negative correlation, suggesting that as interest rates increase, unemployment tends to decrease, or
vice versa. Interest and CPI have -0.88 as correlation value. This indicates, as interest rates increase, CPI tends to decrease. COVID
and Employed Person has a correlation of 0.55 which indicates there is a moderate positive correlation between COVID-19 and
employment, which could imply that employment trends were impacted during the COVID-19 period. Interest and COVID have a
correlation of -0.65 which indicates a negative relationship between interest rates and COVID-19. Some factors, like personal loans,
have weaker correlations with other variables, suggesting that they are less directly related to the primary economic factors here.

The correlation matrix in Figure 2 above shows the relationships between personal loans approved with various economic and
social variables, with values indicating the strength and direction of their correlations. The strongest positive correlation is between
CPI and employed person at 0.87, suggesting that as CPI increases, employment also tends to increase. There is also a strong
positive correlation between the year’s and employed number, which is 0.87 indicating a trend of increasing employment over time.
Additionally, Interest shows a strong negative correlation with unemployed (-0.92), suggesting that higher interest rates are
associated with lower unemployment rates. Interest also has a negative relationship with CPI; -0.88, indicating that higher interest
rates tend to coincide with lower CPI values. Other correlations, like those between personal loan and CPI and between employed
person and COVID, are moderate, while some relationships, such as month with other variables, show weaker correlations,
suggesting little direct relationship. The color coding, with blue representing positive correlations and red representing negative
ones, helps visually highlight these relationships.

After checking the correlation between each variable, the loan applied, and loan approved will be fit with Multiple Linear
Regression (MLR) model. MLR can describe relationships between variables by fitting a line to the observed data. Regression
allows you to estimate how a dependent variable changes as the independent variables’ changes. The analysis of the MLR model is
described in Table 5 below.

Table 5. MLR model output for personal loan applied.

Regression Statistic

Multiple R 0.7104

R Square 0.5046

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 226

Adjusted R Square 0.4530

Standard Error 713.2979

Observations 54

ANOVA

Degree of Freedom SS MS F Significance F

Regression 5 24875977.85 4975195.5703 9.7784 1.731E-06

Residual 48 24422108.12 508793.9192

Total 53 49298085.97

Coefficients Standard Error t Stat p-value Lower 95% Upper 95%

Intercept 74988.07 18844.0067 3.9794 0.0002 37099.661 112876.49

Interest -3756.2 786.5869 -4.7754 1.72E-05 -5337.772 -2174.69

CPI 44.753 129.7651 0.3449 0.7317 -216.1565 305.66

Employed -2.5801 0.8533 -3.0236 0.0040 -4.2958 -0.864

Unemployed -20.093 3.9280 -5.1156 5.43E-06 -27.9916 -12.196

COVID -0.0013 0.0007 -1.8308 0.0733 -0.0028 0.0001

According to the results shown in Table 5, the multiple R value or the correlation coefficient, �� of 0.7104 indicates that the
relationship between the loan applied and the other variables such as interest, CPI, employment, unemployment, and COVID-19 is
quite strong, and the R square or coefficient of determination, ��2 gives a value of 0.5046. This value indicates that 50.46% of the
amount of variation in independent variables can be attributed to the number of loans applied, with the remaining 49.54% due to
other, unexplained factors. Although the ��2 is not very strong, we could not make the conclusion that there is no linear relationship
between the loan applied and the independent variables since the results from the ANOVA table, we reject the null hypothesis of no
linear relationship since Significance F value is less than 0.05.

This low percentage might be caused by the ��-value of the coefficients where CPI and COVID have values greater than 0.05,
indicating that these two factors do not have an influence on the loan applied. If the MLR wants to be used as a modelling purpose,
these variables could be removed and variables could be analyzed again in order to increase the R square value. The MLR model
for the accepted loan can be shown in Table 6 below.

Table 6. MLR model output for personal loan approved.

Regression Statistic

Multiple R 0.7198

R Square 0.5180

Adjusted R Square 0.4678

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 227

Standard Error 274.9676

Observations 54

ANOVA

Degree of Freedom SS MS F Significance F

Regression 5 3900962.85 780192.5703 10.3190 9.275E-07

Residual 48 3629144.97 75607.1868

Total 53 7530107.82

Coefficients Standard Error t Stat p-value Lower 95% Upper 95%

Intercept 25649.68 7264.1338 3.5310 0.0009 11044.160 40255.19

Interest -1553.94 303.2196 -5.1248 5.26E-06 -2163.605 -944.28

CPI -106.77 50.0229 -2.1343 0.0379 -207.344 -6.188

Employed 0.2632 0.3289 0.8000 0.4276 -0.3982 0.924

Unemployed -8.0371 1.5142 -5.3079 2.80E-06 -11.0816 -4.992

COVID -0.0007 0.0003 -2.6274 0.0115 -0.0013 -0.0002

The results are quite similar with the previous analysis of the loan applied especially on the Multiple R and R square, but looking
at the ��-value of the coefficients, only employed variable shows value larger than 0.05 which means that this variable does not has
any significant effect on the loan approval. The MLR fitting from Table 5 and Table 6 shows that, even though the �� and ��2
between loan applied and loan approval has quite similar values, the independent variables that affect the dependent variable are
different.

IV. Conclusions

Based on the correlation matrix in Figure 2, it shows that the personal loan applied, and approval does not have a strong relationship
with all the variables because most of the values are close to 0. However, using the MLR, the Multiple R and ANOVA has proved
that there is a linear relationship between personal loan applied and personal loan approved with interest, CPI, employment,
unemployment, and COVID-19. However, for certain variables such as CPI and COVID-19 with personal loan applied and
employment rate with loan approved, it could be confirmed that these variables have no significant relationship with the selected
dependent variables since both correlation matrix and MLR results give similar findings.

Both correlation matrix and MLR have their own advantages. Correlation matrix observes the relationship for each of the variables.
Although the dependent variables (loan applied and loan approved) show no relationship with the independent variables, it gives
information about other relationships such as employment number and CPI, unemployment number and interest rate, have strong
relationships for both personal loans applied, and personal loan approved. MLR allows researchers to assess the model's variation
as well as the relative contribution of each independent variable to overall variance.

In conclusion, there is a linear relationship between parameters such as interest rate, employment, and unemployment with loan
applied, and the number of personal loan applications is unaffected by COVID-19 situations. However, different conditions for
personal loan approval, such as CPI, interest rates, unemployment, and COVID-19 cases, have an impact on bank loan approvals.

INTERNATIONAL JOURNAL OF LATEST TECHNOLOGY IN ENGINEERING,
MANAGEMENT & APPLIED SCIENCE (IJLTEMAS)

ISSN 2278-2540 | DOI: 10.51583/IJLTEMAS | Volume XIV, Issue IX, September 2025

www.ijltemas.in Page 228

Other than that, for both applied and approved personal loans, employment statistics and interest rates have a very significant
relationship, which is similar to unemployment numbers and CPI.

Acknowledgments

We would like to thank Bank Negara Malaysia and Kementerian Kesihatan Malaysia for giving access to the data used in this study.

References

1. Assous, H. F., & Al-Najjar, D. (2021). Consequences of covid-19 on banking sector index: Artificial neural network model.
International Journal of Financial Studies, 9(4). https://doi.org/10.3390/ijfs9040067

2. David G. Kleinbaum, L. L. K. A. N. E. S. R. (2014). Applied Regression Analysis and Other Multivariable Methods (5th ed.).
Cengage Learning. https://books.google.com.my/books?id=v590AgAAQBAJ

3. Forst, R. (2017). Normativity and power: Analyzing social orders of justification. Oxford University Press.
4. Gur, Y. E. (2024). Development and application of machine learning models in US consumer price index forecasting: Analysis

of a hybrid approach. Data Science in Finance and Economics, 4(4), 469-513.
5. Ismail, S., Ezaili Alias, N., Koe, W.-L., Othman, R., & Halim Mahphoth, M. (2018). Empirical Investigation on Attitude towards

Personal Loans Borrowing. KnE Social Sciences, 3(10), 114. https://doi.org/10.18502/kss.v3i10.3123
6. Kamarudin, S. N., Jasni, N. S., Sal, Z., & Mohamed, S. (2021). The Implications of Covid-19 Pandemic on Consumer Burden

in Malaysia: An Analysis. Journal of International Business, Economics and Entrepreneurship, 6(2), 2550–1429.
https://ir.uitm.edu.my/id/eprint/71586/1/71586.pdf

7. Kumar Tiwari, K., & Bapat, H. B. (2020). Determinants of Loan Delinquency in Personal Loan. International Journal of
Management (IJM, 11(11), 2566–2575. https://doi.org/10.34218/IJM.11.11.2020.241

8. Kung, M. H., Chang, C. C., Hsiao, Y. J., Lo, W. C., & Chang, B. J. (2024). Banks can help? Evidence in the speed of lending for
COVID-19 personal relief loans and financial inclusion. Pacific-Basin Finance Journal, 86, 102448.

9. Najaf, K., Subramaniam, R. K., & Atayah, O. F. (2022). Understanding the implications of FinTech Peer-to-Peer (P2P) lending
during the COVID-19 pandemic. Journal of Sustainable Finance and Investment, 12(1), 87–102.
https://doi.org/10.1080/20430795.2021.1917225

10. Narvekar, A., & Guha, D. (2021). Bankruptcy prediction using machine learning and an application to the case of the COVID-
19 recession. Data Science in Finance and Economics, 1(2), 180-195.

11. Oindrila Chakraborty. (2022). Inflation and COVID-19 Supply Chain Disruption. Managing Inflation and Supply Chain
Disruptions in the Global Economy, 10–23. https://doi.org/10.4018/978-1-6684-5876-1.ch002

12. Tauber, K. N., & Van Zandweghe, W. (2021). Why Has Durable Goods Spending Been So Strong during the COVID-19
Pandemic? Economic Commentary (Federal Reserve Bank of Cleveland), 1–6. https://doi.org/10.26509/frbc-ec-202116

13. Valdivino, C. X., De Paula, T. M., & Gerhard, F. (2025). The use of e-commerce applications in Latin America: individual and
structural influences during COVID-19. Future Business Journal, 11(1), 200.

14. Zainudin, R., Mahdzan, N. S., & Yeap, M. Y. (2019). Determinants of credit card misuse among Gen Y consumers in urban
Malaysia. International Journal of Bank Marketing, 37(5), 1350–1370. https://doi.org/10.1108/IJBM-08-2018-0215