^{1}

^{2}

^{3}

^{1}

Background:
Economic theory suggests that monetary policy through interest rates affects bank profitability. There is limited empirical evidence on the relationship between monetary policy and profitability of commercial banks in Uganda. **Objective: **This study seeks to examine the effect of monetary policy on the profitability of commercial banks in Uganda. **Methodology:** The study adopts a causal relationship research design. Data, covering 9 years from 2010-2018, was collected from all the registered commercial banks which were in operation over the study period. Various monetary policy variables are included in the empirical model as predictor variables. Return on
A
ssets is used as a measure of bank profitability. A dynamic two-step System Generalized Method of Moments panel estimator is applied to estimate the empirical model. **Findings:** Estimates show that monetary policy in terms of its link to the lending rate
has a significant causal effect on Return on Assets, suggesting that interest rate changes predict bank profitability of commercial banks in Uganda.
Further, results
show that a rise in core inflation has a significant negative causal effect on
the banks’ profitability and that there is a significant lagged effect of Return on Assets.
The 91-day treasury bill rate
and money supply were insignificant in predicting bank profitability. **Originality:** Unlike previous related studies which have focused on major advanced economies and a limited number of studies which have considered only a few developing countries like Nigeria and Kenya, the current study provides empirical evidence on the link between monetary policy and commercial bank profitability in Uganda. **Practical Implications:** Policy makers in the financial sector may use the study results as a basis of implementation of appropriate monetary policy actions that enhance the profitability of Uganda’s commercial banks. For instance, the central
bank should promote low and stable core inflation in order to enhance bank profitability, and should ensure that the monetary policy transmission to interest rates is efficient.

The link between monetary policy and bank profitability has gained prominence of recent, particularly after the financial crisis of 2007. Concerns have emerged that the low interest rate monetary policy stance in the Euro area could be affecting bank profitability [

Monetary policy affects bank profitability by influencing the interest rate [

The overall profitability of the Ugandan banking sector, which comprises of 24 commercial banks, has declined over the years to 2018 [

Literature documents several studies that have looked at the effect of interest rates on bank profitability in many countries. For instance, [

Despite the importance of profits for the health and soundness of commercial banks and in enhancing their role in Uganda’s economic growth, indicators show that the average profitability of commercial banks has declined steadily. A number of reforms related to bank to monetary policy have been implemented by government and BOU, which could affect the income earned by banks. For example, in July 2011, Bank of Uganda changed its monetary policy framework from monetary targeting to inflation targeting (IT) lite . In addition, BOU has steadily reduced the central bank rate (CBR) from 23 percent in 2011 to 9.5 percent in 2018 as the banking sector battled high non-performing loans, which ought to have enhanced bank profitability [

The objective of this study is to investigate the effect of monetary policy on profitability of commercial banks in Uganda.

The study seeks to test the following five (5) hypotheses:

H_{01}: Interbank rate has no effect on bank profitability in Uganda.

H_{02}: Treasury bill rate has no effect on bank profitability in Uganda.

H_{03}: Lending interest rate has no effect on bank profitability in Uganda.

H_{04}: Money supply has no effect on bank profitability in Uganda.

H_{05}: Core inflation rate has no effect on bank profitability in Uganda.

In terms of content, the study investigates the influence of monetary policy on profitability of commercial banks in Uganda. Assessing the effect macroeconomic conditions, managerial/operating efficiency and others which may affect bank profitability has not been within the scope of this study. In terms of geographical scope, the study focuses on the commercial banking sector in Uganda. Commercial banks were studied because they link well with the operationalization of the research problem under study and they are key in the transmission of and could be affected by monetary policy. In terms of time scope, the study covered a period of nine years from 2010 to 2018. This period was selected based on data availability across the units (banks) being studied. This period also covers the change in monetary policy framework in Uganda to inflation targeting in the year 2011.

Accordingly, as shown in

Monetary policy is the framework used by the Central Bank to regulate the

circulation of money, interest rates and credit in order to achieve broad economic objectives [

Profit is the driving force of every firm and the main indicator of a firm’s performance and in addition, banks are special types of firms, engaged in mobilizing deposits and lending [

There are three ratios that are typically used to measure the profitability of banks in empirical studies; return on assets (ROA), return on equity (ROE) and net interest margin (NIM) [

Monetary policy rates (Central Bank Rate and Interbank Rate) and bank profitability: The central bank rate is a key monetary policy variable that a central bank sets as a benchmark for all interest rates in an economy, in an inflation targeting regime [

However, these studies focused on major advanced economies and their findings may not be applicable to a developing market like Uganda. In Kenya, [

Lending rates and bank profitability: A study by [

Treasury bill rate and bank profitability: Treasury bills are particularly important to, and are also popular with commercial banks. Moreover, treasury bills count as liquid assets of commercial banks while at the same time earning handsome interest rate for the holders. Treasury bills dominate the money market in Uganda, accounting for the largest portion of all government domestic debt [

Inflation and bank profitability: Empirical literature has documented inflation to be another important determinant of banking performance. However, the findings of the relationship between inflation and bank profitability are mixed. [

Money Supply and bank profitability: A study by [

From the empirical literature reviewed, we note that several studies have investigated the link between monetary policy and bank performance by considering various predictor variables, including treasury bill rate, interbank rate, lending rate, money supply, bank capital and assets. Evidence reveals mixed findings, and there is limited empirical evidence on Uganda. This study contributes to the available literature on the related studies by investigating the role of monetary policy on commercial bank profitability in Uganda.

The study adopts a causal relationship research design. This quantitative approach was chosen because the study used data for individual units (banks) over a period of time. Panel data based multivariate regression analytical procedures are suitable to estimate the effect of common policies and interventions that cut across the units being studied [

At the time of this study, there were a total of 24 registered commercial banks. These 24 commercial banks formed the study population. Out of the 24 commercial banks, 20 of these have been studied. The sample size considered in this study was therefore 20 commercial banks. This sample was studied because the banks within this sample were the commercial banks registered and were operating over the study period. Banks which opened after 2010 or were closed before 2018 were excluded because they had insufficient data.

As suggested by [

The study used secondary data compiled from published annual financial statements of 20 commercial banks, which constituted the unit of observation for the study, as well as data from the Uganda Bureau of Statistics (UBOS) and reports from Bank of Uganda [

The study relied on related studies by previous scholars and economic theory to identify the independent variables and the expected signs of their respective coefficients. Modifications were made where necessary to suit the study context. Data availability was also put into consideration.

Return on Assets (ROA) was used as a measure of bank profitability, that is, the dependent variable was ROA. ROA was defined as bank profit before tax divided by the total assets, which is in line with Bank of Uganda regulations. It indicates management’s ability to utilize banks resources to make profits [

Various monetary policy variables were included in the empirical model as independent variables. These variables include: natural logarithm of money supply (ms), the 91-day treasury bill rate (tbr), 7-day interbank rate (ibr), core inflation rate (coreinf) and weighted average lending rate (lr). In line with some existing studies (for instance [

To guard against the possibility of under fitting the empirical model, the study included control variables in the model, which included: bank total assets (ta), bank capital ratio (capitalr), bank loans (loans) and bank’s holdings of government securities (securities). Empirical results on the bearing of bank assets on bank profitability are mixed and therefore, we predict an indeterminate link between bank assets and bank profitability. On the other hand, some studies support a positive link between capitalization, securities, loans and bank profitability [

The model variables and their respective expected signs are summarized in the table below.

To test the relationship between monetary policy and bank profitability, the study formulated a linear regression model with dynamic specification, considering the dynamic nature of bank variables and the tendency for bank profitability to be serially correlated [

y i t = δ + θ y i t − 1 + ∑ l = 1 m β l Z l t + ∑ j = 1 n α j X j t + v i + u i t (1)

where: y is the variable under study (i.e. bank profitability), θ y i t − 1 is the lagged dependent variable, Z represents the monetary policy variables, X represents control variables, δ is a constant term, v i is the unobserved bank-specific effect, u i t is the idiosyncratic error and subscript t is the time indicator.

In a typical linear dynamic equation such as (1), y i t is a function of μ i . It immediately follows that y i t − 1 is also a function of μ i . Additionally, there are some regressors in X_{jt} (for instance banks total assets) which are endogenously determined). The endogeneity of some regress or in (1) as well the correlation between the lagged dependent variable and the time-invariant country-specific

Variable | Definition | Notation | Expected sign | Reference |
---|---|---|---|---|

Dependent Variable | ||||

Return on Assets | Bank profit before tax divided by the total assets | roa | −/+ | [ |

Independent variables | ||||

7 Day Interbank Rate | Weighted Average Rate at which banks borrow in the interbank market | ibr | + | [ |

91 Day Treasury Bill Rate | Weighted Average Rate for treasury Bills of 91 days tenor. | tb | + | [ |

Money supply | Money supply M2 (base money + shilling deposits) | ms | − | [ |

Lending Rate | Weighted Average rate for bank loans. | lr | + | [ |

Core Inflation | Percentage change in CPI | coreinf | − | [ |

Control Variables | ||||

Total Assets | Total assets of each bank in the sample. | ta | + | [ |

Capital/Assets ratio | Total capital as a share of assets of each bank in the sample. | capitalr | + | [ |

Total Loans | Total loans of each bank in the sample. | loans | + | [ |

Total Securities | Total government securities held by each bank in the sample. | securities | + | [ |

Source: Compiled by the authors.

effects renders the OLS estimator biased and inconsistent. We thus estimate the empirical model by the [

Consider the following generalised linear dynamic panel model:

y i , t = ρ y i , t − 1 + x ′ i , t β + μ i , t + υ i , t ; υ i , t ~ i i d ( 0 , δ υ 2 ) (2)

where y i , t is the depedent variable, y i , t − 1 is the first lag of the dependent variable, μ i represents the bank-specific effects, γ t is the time dummy that captures uncertain shocks, υ i , t is the idiosyncratic error term and x ′ i , t is the vector of predictor variables. In (2), the strict exogeneity assumption of static panel models such as Fixed Effects (RE), Random Effects (FE) and Between Effects (BE) is violated, in a sense that one of the regressors, the lagged dependent variable in is correlated with the past values of the idiosyncratic error term. In other words,

E ( υ i , t | μ i , t , X , y i , s − 1 , ∀ s = 1 , ⋯ , T ) ≠ 0 (3)

In effect, the weaker condition of zero contemporaneous correlation of the regressors with the composite error term ( μ i , t + υ i , t ) is also violated. The composite error terms are also serially correlated due to time-invariant panel specific unobserved effect. The omission of the unobserved effect in the X-matrix breeds another problem of endogeneity. The SGMM addresses all these problems by use of Instrumental variable (IV) of some form.

First consider the DGMM. The DGMM uses the first difference in (2) to eliminate the unobserved effect as follows:

Δ y i , t = Δ ρ y i , t − 1 + Δ x ′ i , t β + Δ ε i , t (4)

where ε i , t = ( μ i , t + υ i , t ) . First differencing however results in a negative correlation between the differenced, lagged dependent variable and the differenced idiosyncratic error term. There is thus still a need to use an IV estimation strategy. IV estimators have been earlier proposed by [

E ( y i , t − s Δ ε i , t ) = 0 , ∀ i , t and s = 2 , ⋯ , ∞ (5)

The moment conditions use the properties of the instruments: y i , t − s ; s ≥ 2 to be uncorrelated with the future errors ε i , t and ε i , t − 1 . An increasing number of of moment conditions is obtained for t = 3 , 4 , ⋯ , T . We then define the (T − 2) × 1 vector:

Δ ε i = [ ( ε i , 3 − ε i , 2 ) , ⋯ , ( ε i , T − ε i , T − 1 ) ] ′ (6)

and a (T − 2) × (T − 2) matrix of instruments as:

Z ′ i = [ y i , 1 y i , 1 y i , 1 ⋯ y i , 1 0 y i , 2 y i , 2 ⋯ y i , 2 0 0 y i , 3 ⋯ y i , 3 ⋮ ⋮ ⋮ ⋱ ⋮ 0 0 0 ⋯ y i , T − 2 ] (7)

Essentially, the past levels of the dependent variable act as instruments for the current first differences of the dependent variable. Also, the exogenous regressors are included in the model as additional instruments, and the additional moment conditions can be formulated such that:

E ( X i , s ε i , t ) = 0 ; ∀ s , t ; ( X i , s Δ ε i , t ) = 0 (8)

for strictly exogenous regressors, and;

E ( X i , s ε i , t ) = 0 ; ∀ s ≤ t ; ( X i , t − j Δ ε i t ) = 0 ; j = 1 , 2 , ⋯ , t − 1 (9)

for predetermined regressors. We can stack these moments up and then apply GMM, which removes the endogeneity bias and omitted variable bias arising from presence of endogenous regressors and omitted variable in the empirical model.

The S-GMM undertakes the D-GMM and augments it through the introduction of additional assumptions which generate an additional set of moment conditions to leverage. The additional assumption is that:

E [ Δ y i , t − s ( μ i , t + υ i , t ) ] = 0 ; ∀ i , t and s = 1 , 2 , ⋯ , ∞ (10)

and the set of instruments to y is composed of blocks that look like:

Z ′ i = [ 0 Δ y i , 2 0 0 ⋯ 0 0 Δ y i , 3 0 ⋯ 0 0 0 Δ y i , 4 ⋯ ⋮ ⋮ ⋮ ⋮ ⋱ ] (11)

and only one lag of y is used for each period as instrumenting variable. The S-GMM requires that the lagged changes in the dependent variable are valid instruments in the level equation. If all the assumptions hold, the S-GMM achieves greater efficiency than D-GMM.

In consideration of the models explained in (1)-(11), we specify an empirical model in form of a dynamic panel model that controls for endogeneity bias and time effects as follows:

r o a i t = δ + θ 1 r o a i t − 1 + α 1 i b r i t + α 2 t b r i t + α 3 l o g m s i t + α 4 l r i t + α 5 c o r e i n f i t + α 6 l o g t a i t + α 7 c a p i t a l r i t + α 8 l o g l o a n s i t + α 9 l o g s e c i t + v i + γ + u i t (12)

where: r o a i t is the return on assets (bank profitability),

δ is a constant term,

θ 1 r o a i t − 1 is the coefficient of the lagged dependent variable,

α 1 i b r is the coefficient of 7-day interbank rate,

α 2 t b r is the coefficient of 91-day treasury bill rate,

α 3 l o g m s is the coefficient of natural logarithm of money supply,

α 4 l r is the coefficient of the lending interest rate,

α 5 c o r e i n f is the coefficient of core inflation rate,

α 6 l o g t a is the coefficient of natural logarithm of bank total assets,

α 7 c a p i t a l r is the coefficient of capital ratio of each bank,

α 8 l o g l o a n s is the coefficient of natural logarithm of bank total loans,

α 9 l o g s e c is the coefficient of natural logarithm of bank’s holdings of government securities.

v i is the unobserved bank-specific effect,

γ is a time dummy that captures shock and time effects, and,

u i t is the idiosyncratic error term.

The Data was processed and analyzed using STATA statistical package, version 14.

First, the data were cleaned by checking for missing values and outliers. Descriptive statistics were summarized to provide a general description of the data characteristics. This helped to ensure that the data was good for estimation, otherwise it would produce misleading results. In particular, a summary of the mean, minimum, maximum and standard deviation values was computed.

Panel regression-based diagnostic tests were conducted to ensure that the data behaves well, and that estimates that are robust are reported. These tests included: panel unit root tests, panel cointegration tests, multicollinearity tests, endogeneity test serial correlation tests and test for validity of instruments.

Given the dynamic specification of the model in this panel study, estimators like random effect (RE), OLS and fixed effect (FE) could not apply, because these estimators would yield biased and inconsistent estimates. One potential problem is the dynamic effects of bank profitability [

This study addressed the above problems by implementing a two-step dynamic System Generalized Method of Moments (S-GMM) panel estimator, which adjusts for the endogeneity bias and corrects for omitted variable bias that is usually due to time-invariant heterogeneity effect across banks. S-GMM is more efficient than the one step Difference GMM (DGMM) estimator developed by [

As proposed by [

The hypotheses in this study were tested as follows:

H_{01}: Interbank rate has no effect on bank profitability in Uganda: This hypothesis was tested by looking at the p-value associated with a ^ 1 in Equation (2). The null hypothesis will not be rejected if the associated p-value > 0.05.

H_{02}: Treasury bill rate has no effect on bank profitability in Uganda: This hypothesis was tested by looking at the p-value associated with a ^ 2 in Equation (2). The null hypothesis will not be rejected if the associated p-value > 0.05.

H_{03}: Lending rate has no effect on bank profitability in Uganda: This hypothesis was tested by looking at the p-value associated with a ^ 4 in Equation (2). The null hypothesis will not be rejected if the associated p-value > 0.05.

H_{04}: Money Supply has no effect on bank profitability in Uganda: This hypothesis was tested by looking at the p-value associated with a ^ 3 in Equation (2). The null hypothesis will not be rejected if the associated p-value > 0.05.

H_{05}: Core Inflation rate has no effect on bank profitability in Uganda: This hypothesis was tested by looking at the p-value associated with a ^ 5 in Equation (2). The null hypothesis will not be rejected if the associated p-value > 0.05.

Return on Assets (roa). This is the dependent variable under study. The descriptive statistics in

Explanatory Variables | Obs N = 20, T = 9 | Mean | Std. Dev. | Min | Max |
---|---|---|---|---|---|

return on assets | 180 | 0.985 | 3.439 | −12.758 | 6.458 |

lending rate | 180 | 22.449 | 2.335 | 19.714 | 26.706 |

91-day treasury bill rate | 180 | 12.716 | 4.968 | 7.967 | 22.890 |

7-day interbank rate | 180 | 13.240 | 5.844 | 6.280 | 27.420 |

core inflation | 180 | 6.335 | 5.891 | 1.970 | 22.341 |

money supply | 180 | 11,416.39 | 3104.0 | 7397.649 | 16,621.90 |

total assets | 180 | 910.40 | 1023.6 | 12.840 | 5423.371 |

total loans | 180 | 430.81 | 491.65 | 3.3 | 2612.6 |

total government securities | 180 | 193.65 | 225.9 | 0.5 | 960.1 |

capital ratio | 180 | 19.435 | 10.479 | 6.1 | 67 |

Source: Authors’ computations and compilation based on raw data.

profitable while others are loss making, which suggests high variability of the data.

Lending rate (lr): The results in

91-day Treasury bill rate (tbr): As shown in

Core Inflation (coreinf): The mean value of “coreinf” over the 9-year period of study was approximately 6.34%, its minimum value was approximately 1.97% recorded in 2010 and its maximum value was approximately 22.34% recorded in 2011. The large variance indicates that core inflation varies widely in some years.

Money supply (ms): The descriptive statistics in

Total Assets (ta): The mean value of “ta” over the 9-year period of study was approximately UGX 910.40 billion, its minimum value was approximately UGX 12.84031 billion in the year 2010 and its maximum value was approximately UGX 5423.371 billion in 2017. The big variance is because some banks are very small in terms of assets while others are very big, with four banks accounting of 50% of the total industry assets [

Total Loans (loans): The mean value of “loans” over the 9-year period of study was approximately UGX 430.8 billion, its minimum value was approximately UGX 3.3 billion in the year 2010 and its maximum value was approximately UGX 2612.6 billion in 2018. The large variance points to disparity in bank’s loan books and business models.

Total Securities (securities): The mean value of “securities” over the 9-year period of study was approximately UGX 193.6 billion, its minimum value was approximately UGX 0.5 billion in the year 2011 and its maximum value was approximately UGX 960.1 billion in 2017. The large variance indicates the difference in bank’s business models with some banks choosing to invest more in securities.

Capital ratio (capitalr): The mean value of “capital” over the 9-year period of study was approximately 19.43%, its minimum value was approximately 6.1% and its maximum value was approximately 67%. The variation in capital indicates differences in funding structures among banks, with risk averse banks having more capital, while loss making banks and risk taking banks having less capital as a share to total assets.

To test for multicollinearity in the model, the study first generates a correlation matrix between the independent variables and then estimates the variance inflation factor (VIF) for each of the independent variables. The results are indicated

[

To further establish which of these variables may cause a multicollinearity problem in the regression model, the study run the variance inflation factor (VIF) for each independent variable in the empirical model.

As indicated in

1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | |
---|---|---|---|---|---|---|---|---|---|

1 | 1 | ||||||||

2 | 0.762** | 1 | |||||||

3 | 0.809** | 0.942** | 1 | ||||||

4 | 0.717** | 0.726** | 0.920*** | 1 | |||||

5 | −0.413** | −0.175** | −0.274** | −0.465** | 1 | ||||

6 | −0.201** | −0.034* | −0.058 | −0.123** | 0.279** | 1 | |||

7 | 0.025 | 0.023 | 0.024 | −0.008** | 0.071 | 0.210** | 1 | ||

8 | 0.013 | −0.012 | −0.022 | −0.076* | 0.206 | 0.808** | −0.218** | 1 | |

9 | −0.531** | −0.065** | −0.111 | −0.185** | 0.304** | 0.803** | −0.178** | 0.828** | 1 |

Explanatory variables | VIF | 1/VIF |
---|---|---|

interbank rate | 115.93 | 0.0086 |

logarithm of total assets | 44.39 | 0.0225 |

treasury bill rate | 32.63 | 0.0306 |

core inflation | 31.02 | 0.0322 |

logarithm of total loans | 24.18 | 0.0413 |

logarithm of total securities | 10.18 | 0.0982 |

lending rate | 7.26 | 0.1378 |

logarithm of money supply | 5.75 | 0.1738 |

First lag of return on assets | 2.05 | 0.4882 |

Capital ratio | 1.20 | 0.8347 |

Mean VIF | 27.46 |

Explanatory variables | VIF | 1/VIF |
---|---|---|

interbank rate | 6.84 | 0.1463 |

treasury bill rate | 5.33 | 0.1875 |

core inflation | 4.86 | 0.2056 |

logarithm of money supply | 4.21 | 0.2372 |

logarithm of total assets | 1.81 | 0.5511 |

first laog of return on assets | 1.75 | 0.5707 |

capital ratio | 1.18 | 0.5844 |

Mean VIF | 3.71 |

From

The study conducted stationarity tests on model variables, using the [

The panel unit root test results in

The study employed the [

The ADF t-statistic of the [

Model Variables | Variable in levels | Variable in first difference | Order of integration | ||
---|---|---|---|---|---|

W-t bar Statistic | p-value | W-t-bar Statistic | p-value | ||

return on assets | −1.0368 | 0.1499 | −6.6057 *** | 0.0000 | I (1) |

lending interest rate | −2.2813** | 0.0113 | - | - | I (0) |

treasury bill rate | −7.5633*** | 0.0000 | - | - | I (0) |

core inflation | −11.1917*** | 0.0000 | - | - | I (0) |

logarithm of money supply | −1.0459 | 0.1478 | −5.3316*** | 0.0000 | I (1) |

logarithm of total assets | 0.4135 | 0.1225 | −0.012*** | 0.0000 | I (1) |

capital ratio | −1.689** | 0.035 | I (0) |

Source: Authors’ compilation; **means that the estimate is statistically significant at 5 percent level of significance; ***means that the estimate is statistically significant at 1 percent level of significance.

t-Statistic | Prob. | |
---|---|---|

ADF | −5.044 | 0.0000 |

Residual variance | 0.478 | |

HAC variance | 0.123 |

Source: Generated and compiled by the authors.

evidence of presence of long run equilibrium relationships between return on assets (roa) and its determinants.

Following previous studies for instnace [

At 5 percent level of significance, the test does not reject the null hypotheses that variables: “lr”, “ibr”, “coreinf”, “logms” and “logcapital” are exogenous. on the other hand, “logta” has a p value < 0.05, so we reject the null hypothesis and conclude that the variable is endogenous at 5 percent level of significance. This suggests that it is an endogenous regressor in the model.

As previously stated, the study estimated the empirical model using System GMM technique following related studies (such as [

We interpret the estimates in

H_{01}: Interbank rate has no effect on bank profitability in Uganda:

The variable “interbank rate” was removed from the empirical model because it had a high variance inflation factor. Its inclusion in the final model would have caused multicollinarity problem in the model. It was therefore not possible to test the hypothesis on its effect on return on assets in the final model.

H_{02}:Treasury bill rate has no effect on bank profitability in Uganda:

The regression results in

Null hypothesis being tested | Sargan | P > Chi.Sq |
---|---|---|

Ho: “lr” is exogenous | 1.770 | 0.1834 |

Ho: “tbr” is exogenous | 3.163* | 0.0753 |

Ho: “coreinf” is exogenous | 0.126 | 0.7226 |

Ho: “logms” is exogenous | 0.278 | 0.5979 |

Ho: “logta” is exogenous | 6.576** | 0.0103 |

Ho: “capitalr” is exogenous | 0.094 | 0.7592 |

Source: Compiled by the author; *means that the estimate is statistically significant at 10 percent level of significance; **means that the estimate is statistically significant at 5 percent level of significance.

Dependent Variable: Return on Assets Estimation method: two-step System GMM | |||
---|---|---|---|

Independent Variables | Coef. | Corrected/robust Std. Err. | p-value |

First lag of return on assets | 0.2780*** | 0.1173 | 0.009 |

Treasury bill rate | 0.0532 | 0.4725 | 0.260 |

Lending interest rate | 0.3666** | 0.1782 | 0.040 |

Core inflation | −0.2490** | 0.1271 | 0.050 |

Logarithm of money supply | −0.3233 | 1.282 | 0.801 |

Logarithm of total assets | 1.3917*** | 0.4373 | 0.001 |

Capital ratio | 0.0974*** | 0.0334 | 0.004 |

Constant | −4.378 | 14.330 | 0.309 |

Diagnostic tests | |||

Arellano-Bond test for AR(1) in first differences: z = −2.87***; Pr > z = 0.004 | |||

Arellano-Bond test for AR(2) in first differences: z = −1.51; Pr > z = 0.132 | |||

Sargan test of overid. Restrictions: Chi-Sq. = 108.55*; Prob > chi2 = 0.051 | |||

Hansen test of overid. Restrictions: Chi-Sq. = 39.85; Prob > chi2 = 0.192 | |||

Difference-in-Hansen (iv): Chi-Sq. = 12.78; Prob > chi2 = 0.991 | |||

Difference-in-Hansen(gmm): Chi-Sq. = 9.40; Prob > chi2 = 0.152 | |||

Wald Chi-square = 112.87***; Prob > chi2 = 0.000 | |||

Standard instruments for first differences equation: D. (logcapital logloans logsecurities 2010b. year 2011. year 2012. year 2013. year 2014. year 2015. year 2016. year 2017. year 2018. year) GMM-type Standard instruments for first differences: L (1/8). logta Standard instruments for levels equation: logcapital logloans logsecurities 2010b. year 2011. year 2012. year 2013. year 2014. year 2015. year 2016. year 2017. year 2018. year GMM-type for levels equation: D.logta |

Source: Compiled by the authors. *means that the estimate is statistically significant at 10 percent level of significance; **means that the estimate is statistically significant at 5 percent level of significance; ***means that the estimate is statistically significant at 1 percent level of significance

on returns on assets of commercial banks in Uganda.

H_{03}: Lending interest rate has no effect on bank profitability in Uganda:

The regression estimates in

H_{04}: Money Supply has no effect on bank profitability in Uganda:

The estimated coefficient on the logarithm of money supply is negative and statistically insignificant at 5 percent level of significance (coef. = −0.323; p = 0.801). Thus we accept the hypothesis, which suggests that changes in money supply do not have a significant causal effect on returns on assets of commercial banks in Uganda.

H_{05}: Core Inflation rate has no effect on bank profitability in Uganda:

The regression estimates from

The study checks for the robustness of the regression estimates by performing the following relevant post estimation diagnostic tests after model estimation.

Wald Chi-square statistic: This statistic tests the null hypothesis that all the parameters of the model in Equation (4) are simultaneously equal to zero. The estimated Wald Chi-square statistic of 112.87 with p = 0.000) is statistically significant at 5 percent level. Thus we reject the null hypothesis and conclude that the model is well specified and that monetary policy, the included control variables as well as the lag of return on assets, have a combined effect that is non-zero(i.e. that is statistically significant) on bank profitability in Uganda.

AR (1) and AR (2) tests: As shown in

Hansen Test: This test verifies the validity of the instruments [

Treasury Bill rate and bank profitability: Model estimates showed that there is a positive but statistically insignificant relationship between the 91-day treasury bill rate and profitability of banks in Uganda, implying that changes in the 91-day treasury bill rate are not important predictors of bank profitability. Our study results are consistent with the findings from other related studies on Uganda, for instance, [

Core Inflation and bank profitability: Model estimates showed that core inflation exerts a negative and significant causal effect on bank profitability. This result means that as core inflation rises, bank’s return on assets reduces in Uganda’s commercial banks. This finding is consistent with monetary theory which postulates that rising inflation reduces real interest income, leads to a deterioration in asset quality [

Money supply and bank profitability: The estimated coefficient on money supply variable was statistically insignificant at 5 per cent level. This suggests that changes in money supply do not have a significant causal effect on returns on assets of commercial banks in Uganda. This finding is consistent with monetary theory, whereby as countries move from monetary targeting to inflation targeting, the interest channel of monetary policy becomes more effective relative to the other channels [

Lending interest rate: Estimates showed a positive and statistically significant coefficient on lending interest rates. This suggests that variations in lending interest rates have a significant positive causal effect on return on assets in Uganda’s commercial banks. The implication of this result is that changes in monetary policy transmitted to lending interest rates influence bank profitability and this suggests the presence of the interest rate channel of the monetary policy transmission mechanism. In Uganda, all loans are on variable interest rates and thus monetary policy changes are passed through to customers. For example, BOU reduced the policy rate from 23 percent in December 2011 to 9.5 percent by December 2018, and correspondingly, the weighted average lending interest rate reduced from 29.5 percent to 19.5 percent [

Bank profitability persistence: The estimated coefficient on the first lag of the dependent variable (coef. = 0.2780; p = 0.009) is positive and statistically significant at 5 percent level of significance. This conﬁrms the dynamic character of the panel model and also indicates that return on assets in the past year has a significant positive causal effect on return on assets of the banks in the current period, and this suggests that there is persistence of profitability among Ugandan banks. The existence of persistence of bank profits indicate that there are some impediments to market competition, which allows abnormal profits to persist over time among a few banks, while convergence has been slow. There is thus a need for policymakers to implement measures to enhance market competition and for small banks to pursue profitability first rather than growth.

Overall, the study finds that monetary policy as measured by variables such as lending interest rates and core inflation are significant predictors of commercial bank profitability in Uganda, the former having a positive influence while the latter having a negative influence. Findings from this study further show that the other monetary policy variables such as money supply and treasury bill rate do not influence commercial bank profitability in Uganda. The interbank rate monetary policy variable was removed from the empirical model because its inclusion would cause high multicollinearity. The study results suggest that monetary policy through its link with lending interest rates affects banks’ profitability and net interest income as it increases bank interest margins and returns from maturity transformation. It also implies that interest rate pass through to lending rates may be working. However, the pass through of policy rate through the 91-day treasury bill rate appears to have weakened. A more efficient monetary policy transmission will ensure better interest rates pass through to banks and enhance profitability. The findings also show that an increase in inflation will also negatively affect the cost and revenue functions of the bank and therefore low and stable inflation is key for bank profitability.

The following are the key policy recommendations the study derives from the analysis:

1) The finding that the 91-day treasury bill rate is not significant in influencing the profitability of banks suggests that the central bank should conduct regular studies to identify other market rates that are effective in transmission of monetary policy.

2) The finding that variations in the lending interest rates affect bank profitability significantly suggests that, to ensure a sustainable strong banking sector, the central bank should ensure that monetary policy transmission is efficient in line with macroeconomic conditions. There is also need for the central bank to monitor the micro-dynamics of individual bank behavior and continuously assess and enhance the efficacy of the interest rate pass through to the lending channel of monetary policy transmission mechanism. This will improve the availability of credit for corporate and private investment and enhance bank profitability.

3) Rising inflation constraints bank profitability and therefore, to ensure and maintain a sound financial stance in Uganda, the central bank and government should strive to achieve and maintain lower levels of core inflation through credible monetary and fiscal policy interventions.

We recognize the fact that the overall effect of monetary policy on bank profits will also depend on the impact of monetary policy on macroeconomic conditions. In particular, it will crucially hinge on the efficacy of monetary policy in boosting aggregate demand. Assessing this relationship has been beyond the scope of this study. In addition, this study considers commercial banks and thus the results from the study may not be used to assess the relationship ship between monetary policy and profitability in other financial institutions such as pension funds. In terms of interest rate pass through, the study only looked at the 91-day treasury bill rate. A look at other money market rates may yield different results. Future studies could also extend the analysis by looking at the effect of interest rate increases on banks by category i.e. small, medium and big banks, as well as other financial institutions such as pension funds and microfinance institutions.

The authors declare no conflicts of interest regarding the publication of this paper.

Mbabazize, R.N., Turyareeba, D., Ainomugisha, P. and Rumanzi, P. (2020) Monetary Policy and Profitability of Commercial Banks in Uganda. Open Journal of Applied Sciences, 10, 625-653. https://doi.org/10.4236/ojapps.2020.1010044