Page 3451
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
DOI:
https://doi.org/10.51244/IJRSI.2025.1210000300
This study aimed to investigate the impact of renewable energy usage on Kenya’s manufacturing and industrial
sectors. It emphasizes the sector’s transition toward green manufacturing and industrial growth. The study made
use of annual time series data for the period 1985-2023. It employed the ARDL framework, guided by the AIC
criterion and the ECM to capture both short-run and long-run dynamics amongst the variables. The ARDL
findings revealed that the use of renewable energy has immediate positive impacts on manufacturing output and
productivity. The lagged effects emerge over multiple periods, implying the gradual integration of renewable
energies into the production processes. On the other hand, non-renewable energy consumption has delayed
negative effects, thus highlighting the shortcomings that come with the usage of non-renewable energies. Labour
has a strong positive influence on output, whereas gross capital formation shows a long-lag positive impact. This
shows the extended gestation periods of capital-intensive projects. The ECM results confirm that deviations from
long-run equilibrium are corrected at a moderately fast rate of about 52% annually. The F-bounds test found the
existence of a long-run association between renewable energy consumption and manufacturing and industrial
output. These results emphasize the importance of renewable energy uptake, efficient capital deployment and
labour utilization in driving sustainable manufacturing and industrial growth in Kenya. This study provides
policymakers and industry stakeholders with empirical guidance for aligning energy policy, investment and
technological adoption with Kenya’s broader objectives for green industrialization. This will ensure countries
not only grow economically, but also grow sustainably since renewable energies have the capability of taking
care of both the current and future energy needs without fear of depletion, while also ensuring environmental
sustainability. This study, therefore, highlights that renewable energy consumption significantly supports
Kenya’s manufacturing sector growth and recommends aligning industrial energy policy interventions with
green growth and sustainability frameworks such as Vision 2030 and SDG 9 by promoting clean energy
adoption, energy-efficient technologies and renewable energy integration within industrial value chains so as to
enhance competitiveness and low-carbon industrial transformation.
renewable energy consumption, manufacturing and industrial activities, greening the industrial
growth
The manufacturing and industrial sector is critical in modernization and global competitiveness and contributes
approximately 17% of the global GDP. Countries such as China and South Korea have achieved accelerated
sectoral growth through manufacturing and industrialization, which depend heavily on affordable and reliable
energy. (World Bank, 2021).
The manufacturing and industrial sector in Kenya plays a critical role in accelerating industrial production,
employment creation and economic diversification, among others. However, the sector’s growth has for some
Page 3452
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
time now been hindered by unreliable and costly energy supply, mainly from non-renewable sources, thereby
increasing the costs of production. Given the country’s rapid economic expansion and the ever-rising demand
for electricity, the integration of renewable energy sources such as hydroelectric power, solar and wind has
proved to be a strategic approach to ensure energy security, reduce the skyrocketing production costs and also
promote environmentally sustainable manufacturing and industrial development. Despite the investments in
renewables, challenges remain in effectively and efficiently translating renewable energy usage into tangible
manufacturing and industrial productivity. Changes in infrastructure readiness, capital deployment and the rate
of workforce and labour absorption can serve to hinder the sectoral benefits of clean and green energy. Existing
literature, such as Oyeleke and Akinlo (2020) in Nigeria and Jeon (2022) across US states, has clearly shown
the positive impacts of renewable energy usage on the sectoral output, yet statistics from Kenya, particularly
within the manufacturing and industrial sector, remain to be limited. This study sought to bridge this empirical
gap.
Kenya’s manufacturing and industrial sector continues to face frequent power shortages, load-shedding, high
costs of electricity from non-renewable sources and limited integration of renewable energy infrastructure. While
the adoption of renewable energy is increasing, its direct contribution to manufacturing sector output, both
immediately and lagged, remains to be unknown. The understanding of the sectoral impact of renewable energy
usage is important for guiding investment in the country, policy formulation and technological adoption as well
as for achieving Kenya’s long-term industrial and clean green growth aspirations.
This study sought to examine the short-run effects of renewable energy consumption on manufacturing and
industrial sector growth in Kenya and to also provide policy recommendations and interventions for enhancing
manufacturing and industrial growth through renewable energy deployment and adoption.
This study sought to provide policymakers, energy investors and industrial stakeholders with evidence-based
insights into the important role of renewable energy usage in promoting and accelerating the growth of the
manufacturing and industrial sectors in Kenya. By putting figures on both immediate and lagged effects of
renewable energy uptake, this study highlights the timing and magnitude of renewable energy interventions that
are deemed in the optimization of manufacturing and industrial output. It also supports Kenya’s agenda for
energy transition, climate mitigation and preservation and middle-income industrialization by showing how
renewable energy consumption can stabilize manufacturing and industrial output while at the same time limiting
the overdependence on non-renewable energy sources.
Existing studies on energy and manufacturing and industrial growth has largely been focusing on the aggregate
performance of the economy. Few studies have labored to explore the sector-specific impacts of renewable
energy consumption on Kenya’s manufacturing and industrial output using robust econometric techniques such
as the ARDL and the ECM. This study fills this gap by providing a comprehensive analysis of both short-run
and long-run effects, while incorporating lag structures and ECM dynamics, while controlling for labour and
capital inputs and the non-renewable energy consumption.
Consumption of renewable energy is the use of energy generated from natural resources such as sunlight, wind,
rain, geothermal and heat, which are renewable and do not suffer from the fear of depletion (
Guliyev & Tatoglu,
2023)
. Several empirical studies exist to explain the relationship between renewable energy consumption and
the corresponding growth of the manufacturing, service and agricultural sectors.
Studies that are in consonance with the growth hypothesis within the energy-economic growth framework
suggest that renewable energy consumption has a significant positive effect on manufacturing output. An
example is the assertion by Obange et al. (2013) that examined the influence of renewable energy on
manufacturing growth in Kenya. The study found a bi-directional Granger causality between electricity
consumption and manufacturing, observed in both short and long-run periods. Using time series data from 1970
to 2010, the study highlighted the existence of a relationship but did not delve into the magnitude or intensity of
Page 3453
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
these effects. In contrast, this study employed advanced analytical methods to not only confirm the presence of
such relationships but also to analyze the direction, strength and scale of the impacts and interactions Ototo and
Nzai (2022) using annual time series data from 1980-2019 while exploiting the multivariate time series model
and incorporating relevant time-series diagnostics, including cointegration test, to examine the short-run and
long-run effects of renewable and non-renewable energy consumption on manufacturing sector performance in
Kenya, established a feedback positive interaction existing between the usage of renewable energy alternatives
and the performance of the manufacturing and industrial sector. However, the study advocates for the
exploitation of energy-efficient infrastructure to check energy losses within the value chain. The study, apart
from exploiting a less robust analysis technique, did little to establish the magnitudes of the associations. The
ARDL methodology that this study exploited was capable of establishing both short-run and long-run
magnitudes of the suspected associations.
According to Hoang et al. (2020), a study that sought to examine the relationship between renewable energy
usage and industrial performance in the United States of America monthly over the 1981-2018 period by
employing the Granger-causality test, found that a positive and significant co-movement exists between the
usage of renewable energy and industrial performance. The study adopted the usage of fossil-based energy as
well as crude oil prices as control variables. Hoang et al. (2020), however, did little to establish the magnitudes
of the associations. The United States is in a diverse stage of economic advancement and therefore, the study
cannot be replicated in Kenya. This body of knowledge sought to clear the ambiguity regarding the sign,
direction and magnitudes of the interactions.
Kasae (2014), in its effort to establish energy efficiency in Kenya’s manufacturing sector using baseline and
current data in Kenya, established that efficiency in the energy value chain possessed a significant influence on
manufacturing productivity and growth. The study focused on a sample of 70 manufacturing firms from a total
of 735 listed by the Kenya Association of Manufacturers, with a particular emphasis on firms that had previously
conducted energy audits
. The study utilized regression analysis. This implies that energy consumption led to
increased manufacturing growth. However, the study treated energy consumption as an aggregate. According to
Kasae (2014), the study is also not definitive as findings from other companies were not in consonance with
these results.
The study also notes that this finding was inconclusive, as certain companies exhibited weak
correlation coefficients for the variables at both baseline and current levels, indicating the necessity for further
investigation. This study helped clear the ambiguity, besides also incorporating other sectors of the economy,
not just the manufacturing sector.
Onuonga et al. (2008), in its quest to determine the existing relationship between the two key parameters, did an
econometric assessment to determine the kind of association existing between manufacturing output and energy
use in Kenya. The study adopted a trans-logarithmic function to assess total factor specifications involving fuel
substitutions. It utilized data from 1970 to 2005. The findings indicated that manufacturing growth was linked
to an increased use of various inputs, including different types of energy. However, Onuonga et al. (2008) failed
to accurately single out the specific effect on the manufacturing sector given the fact that manufacturing activities
were being influenced by many other factors in the model. This study not only singled out the specific impacts
of renewable energy constructs on the manufacturing sector but also sought to clear the same ambiguity about
the agricultural and service sectors of the economy.
Forkuoh and Li (2015) sought to unravel the nexus between electricity insecurity and the growth of small
manufacturing and industrial plants in Ghana. It sought to determine the impact attributable to power outages
on the effectiveness of the manufacturing plants. It was found that there was a positive relationship between
energy efficiency and usage and the growth of the firms. The study also established that the use of backup
generators during power outages led to increased costs of production, thereby suppressing production. By the
use of mixed methods of study and SPSS output, the study, that solely focused on the Asafo market established
that power availability and reliability affected the performance of the manufacturing plants. The study, however,
is from an economy with different natural and energy resource endowments compared to Kenya. This study was
specific to Kenya and economies with similar resource endowments, economic conditions and political climate
as Kenya.
Page 3454
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Grainger and Zhang (2017) sought to evaluate the influence of electricity shortages on the operational
performance of manufacturing firms in Pakistan. The study employed a cross-sectional regression analysis using
firm-level data, where the natural logarithms of firm revenues and value-added were regressed on indices
measuring the duration and frequency of electricity shortages, while controlling for labour costs, raw material
costs, fixed assets, total electricity costs and sector fixed effects to estimate the impact of power outages on firm
productivity. It considered the performance of 4,500 firms for the period 2010-2011. Electricity shortages were
measured as the number of hours that the power blackouts lasted. The study established that a 10% rise in the
number of outage hours led to a 0.14% decline in firms’ revenues and a 0.36% decline in value-added. This
implied that it positively influenced manufacturing output and growth. However, the study ignored the intensity
of the associations. This study aided in not only determining the presence but will also help determine the sign,
direction and intensities of the associations.
Bowden and Payne (2010) conducted a study focusing on the United States, utilizing time-series data from 1949
to 2006. It made use of the Toda-Yamamoto approach to explore the relationship between renewable energy
consumption and economic growth. Their long-run causality tests indicated no Granger causality between
commercial and industrial renewable energy consumption and real GDP. However, bidirectional Granger
causality was found between commercial and residential non-renewable energy consumption and real GDP.
Additionally, the results showed unidirectional causality from residential renewable energy consumption and
industrial non-renewable energy consumption to real GDP. A key limitation of this study is its exclusive focus
on the United States, which may restrict the applicability of the findings to other regions with differing energy
frameworks. This study sought to bridge this geographical barrier.
While seeking to establish the nexus between renewable energy use and export performance of manufacturing
firms in India between 2011 and 2021, Das and Mahalik (2023), using both the system dynamic and panel
estimation and fixed effects with Driscoll and Kraay standard errors so as to control for cross-sectional
independence, established that a positive impact between renewable energy consumption and manufacturing
productivity in India. This study, despite attempting to establish and aggregate the renewable energy
consumption effects at the sectoral level, is from an economy at a different economic stage compared to Kenya,
hence this study had to be conducted.
Zhang and Ma (2023), in China, while seeking to establish the impact of energy consumption structure
transformation on total firms’ factor productivity using data for the period 2010-2019, found that promotion of
the use of this energy source boosted growth of the sector. The study, though, is from a more advanced stage of
economic development.
For the manufacturing sector growth, this study adopted the correlational research design and employed the
pragmatist research philosophy to use the Solow Swan growth model, the Cobb-Douglas framework to unearth
the relationship between renewable energy consumption and manufacturing sector growth in Kenya.
Nonrenewable energy consumption, gross capital formation and labour, L, were incorporated as control
variables.
The basic Cobb-Douglas function was linearized by logging and therefore the results were interpreted as
percentages and elasticities.
𝐿𝑛(𝑀𝐴𝑁
𝑡
) = 𝐿𝑛𝐴 + 𝛼
1
𝐿𝑛(𝑅𝐸𝐶
𝑡
) + 𝛼
2
𝐿𝑛(𝑁𝑅𝐸𝐶
𝑡
) + 𝛼
3
𝐿𝑛(𝐾
𝑡
) + 𝛼
4
𝐿𝑛(𝐿
𝑡
) + 𝛽
1
𝐿𝑛(𝑅𝐸𝐶
𝑡−
1
) + 𝛽
2
𝐿𝑛(𝑁𝑅𝐸𝐶
𝑡−
1
)
+ 𝛽
3
𝐿𝑛(𝐾
𝑡−
1
) + 𝛽
4
𝐿𝑛(𝐿
𝑡−
1
) + 𝜀
𝑡
………………..……. (3.1)
Where 𝐿𝑛𝑀𝐴𝑁
𝑡
is the log of manufacturing output, at time t, 𝐿𝑛(𝑅𝐸𝐶
𝑡
) is the logged renewable energy
consumption at time t, 𝐿𝑛(𝑁𝑅𝐸𝐶
𝑡
) is logged non-renewable energy consumption at time t, 𝐿𝑛(𝐾
𝑡
) is logged
capital at time t, 𝐿𝑛(𝐿
𝑡
) is logged labour at time t, 𝐿𝑛(𝑅𝐸𝐶
𝑡−
1
) is lagged renewable energy
Page 3455
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
consumption,𝐿𝑛(𝑁𝑅𝐸𝐶
𝑡−
1
) is lagged non-renewable energy consumption,𝐿𝑛(𝐾
𝑡−
1
) is lagged capital, 𝐿𝑛(𝐿
𝑡−
1
)
is lagged labour 𝜀
𝑡
is the error term while 𝐿𝑛𝐴 is logged total factor productivity.
MAN (”0000
00”)
REC
NREC (“0000
00”)
L (“{000000”
)
GCF (“OF_G
DP”) (K)
Mean
486735.3
19791.70
0.137666
13.54270
20.09076
Median
424905.7
16115.00
0.117500
12.73076
19.73131
Maximum
848461.0
46600.00
0.242000
23.18485
25.44904
Minimum
227142.4
6000.000
0.072000
5.341202
15.00382
Std. Dev.
178639.4
10732.72
0.057575
5.556273
2.956138
Skewness
0.455448
0.900081
0.577631(
0.205444
0.087421
Kurtosis
2.110278
2.769333
1.841721
1.682008
2.157324
Jarque-Bera
2.972451
6.038611
4.906443
3.494208
1.357899
Probability
0.226225
0.048835
0.086016
0.174278
0.507149
Sum
21416353
870835.0
6.057312
595.8787
883.9934
Sum Sq. Dev.
1.37E+12
4.95E+09
0.142540
1327.503
375.7664
Observations
44
44
44
44
44
(Source: Author,2025)
Table 4:1 presents the descriptive statistics for the variables used in the study. It covers manufacturing outputs,
renewable energy indicators, gross capital formation and labour. Each variable is interpreted based on its
statistical characteristics and put in the right Kenyan context.
The average output of the manufacturing sector, measured in constant Kenyan shillings, stood at Kshs 487
billion over the study period. The sector exhibited a standard deviation of Kshs 179 billion, with values
ranging between Kshs 227 billion and Kshs 848 billion. This substantial spread reflects the sector's fluctuating
performance, likely influenced by industrial policy changes, access to electricity, global economic shifts and
internal inefficiencies. The moderate right-skewness value of 0.455 and near-normal kurtosis value of 2.11
support the conclusion that manufacturing output has generally trended upward with occasional sharp
increases. The J-B probability of 0.226 confirms the data’s
normality. These trends make sense since Kenya’s manufacturing growth has been uneven, shaped by policy
reforms, energy costs and changing global demand. The overall upward trajectory reflects gradual industrial
recovery and expanding domestic production capacity.
Renewable energy consumption recorded an average of 19791.70 kilowatt-hours, with a standard deviation
of 10732.72. The consumption ranged from 6000- 46600 kilowatt-hours, reflecting the country’s gradual
transition to green energy. The right-skewness of 0.900 and moderate kurtosis of 2.77 imply a growth pattern
where most of the observations are concentrated in earlier, lower-consumption years, with sharp increases in
later periods. The J-B probability of 0.048 confirms the distribution is marginally non-normal, supporting the
choice to log transform this variable in regression analyses. This trend is economically consistent with
Kenya’s growing investment in renewable energy projects, particularly geothermal and hydropower, which
Page 3456
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
have steadily expanded capacity and boosted consumption as part of the national shift toward sustainable
energy use.
Non-renewable energy consumption showed a higher average of 137666.2 kilowatt-hours and a standard
deviation of 57575.04, with values ranging from 72000-242000. This steady growth corresponds with rising
transport and industrial demands for non-renewable energies. The right-skewed distribution of 0.578 and mild
departure from normality, with a J-B probability of 0.086, indicate a consistent upward trend punctuated by years
of accelerated consumption. This pattern is expected as Kenya’s transport and manufacturing sectors still heavily
depend on fossil fuels, leading to sustained growth in non-renewable energy use despite ongoing efforts to
expand cleaner alternatives.
Labour showed a mean of 13542697 with a standard deviation of 5556273. The population grew from 5341202
to 23184849, nearly quadrupling over the study period. The low skewness of 0.205 and normal distribution, as
indicated by a J-B probability of 0.174, reflects steady demographic growth, a crucial determinant of labour
supply in the economy. This pattern can be explained by the fact that Kenya’s steady population growth has
continuously expanded the labour force, providing a vital foundation for sectoral productivity and domestic
market development.
Gross capital formation, expressed as a percentage of GDP, averaged 20.09076%, with a standard deviation of
2.956138%. The values ranged between 15.00382% -25.44904%, indicating sustained investment efforts. The
very low skewness value of 0.087 and strong normality, with a J-B probability of 0.507, suggest a stable
macroeconomic environment that is conducive to infrastructure development and industrial growth. This
outcome can be justified by the fact that consistent gross capital formation reflects Kenya’s ongoing commitment
to infrastructure expansion and industrial investment, which underpin long-term growth and structural
transformation.
The descriptive statistics align with Kenya’s historical trajectory of structural transformation, increasing
investment in renewable energy, demographic expansion and shifting sectoral contributions to GDP. The
presence of skewness and kurtosis in most energy variables confirms the appropriateness of log transformations
to normalize the data for sound econometric modeling. This overall pattern is economically sound as it captures
Kenya’s gradual evolution toward a more diversified and energy-driven economy, where renewable investments,
population growth and sectoral shifts collectively shape the country’s development path.
Null hypothesis: Variable has a unit root Lag length: Automatic based on AIC, maximum lags of 10
ADF
Level
First Difference
CONCLUSION
Variable
Trend & Intercept
Trend & Intercept
MAN
-0.478092 (0.9809)
-5.886987 (0.0001)
I (1)
REC
-2.148084 (0.5053)
-5.148163 (0.0008)
I (1)
NREC
-2.591905 (0.2858)
-6.793060 (0.0000)
I (1)
L
-3.558792 (0.0494)
-2.925557 (0.1665)
I (0)
GCF____OF_GDP
-2.754643 (0.2212)
-5.703648 (0.0002)
I (1)
(Source: Author, 2025)
Table 4.2 shows the results of the ADF stationarity test. In order to ensure the validity of time series regression
Page 3457
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
analysis, particularly in ARDL modelling, it is important to test the stationarity properties of the variables under
study. The ADF test was exploited to examine whether each variable contains a unit root, implying
nonstationarity or whether it is stationary either at level or if it becomes stationary after first differencing. The
test was conducted using the trend and intercept specification, with lag length automatically selected based on
the AIC and a maximum lag of 10.
The results reveal that the dependent variable, that is, manufacturing output is non-stationary at level but become
stationary after first differencing, thus integrated of order one. The same trend is observed for the energy
variables: renewable energy consumption and non-renewable energy consumption, all of which are stationary at
first difference and therefore also classified as I (1). In contrast, labour is found to be stationary at level, implying
it is integrated of order zero. This means labour does not require differencing to achieve stationarity. Gross
capital formation is both non-stationary at level but become stationary after first differencing, confirming that
they are also I(1) variables.
The combination of variables with mixed orders of integration, that is, I(0) and I(1), satisfies the fundamental
requirement for using the ARDL bounds testing approach, which allows for such a mix as long as none of the
variables is integrated of order two, I(2). Since none of the variables is I(2), the dataset is appropriate for ARDL
modelling and further cointegration analysis can proceed confidently using the bounds testing procedure.
The ARDL model automatically selected the lag lengths using the Akaike Information Criterion which is known
in helping strike a good balance between having a well-fitting model and keeping it simple.
.
Dependent Variable: MAN
Method: ARDL
Date: 07/08/25 Time: 18:14
Sample (adjusted): 1985 2023
Included observations: 39 after adjustments
Maximum dependent lags: 1 (Automatic selection)
Model selection method: Akaike info criterion (AIC)
Dynamic regressors (5 lags, automatic): REC NREC L GCF
Fixed regressors: C
Number of models evalulated: 1296
Selected Model: ARDL(1, 4, 1, 0, 5)
Variable
Coefficient
Std. Error
t-Statistic
Prob.*
MAN(-1)
0.477181
0.122378
3.899221
0.0007
REC
0.071733
0.022235
3.226129
0.0037
REC(-1)
0.018555
0.022932
0.809136
0.4267
REC(-2)
0.023068
0.025104
0.918899
0.3677
REC(-3)
0.044265
0.026708
1.657385
0.1110
REC(-4)
0.063231
0.025846
2.446420
0.0225
NREC
-0.038739
0.063556
-0.609526
0.5482
NREC(-1)
-0.113673
0.054764
-2.075683
0.0493
L
0.353834
0.097750
3.619785
0.0014
GCF
0.044694
0.033286
1.342713
0.1925
Page 3458
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
GCF(-1)
0.031500
0.038064
0.827551
0.4164
GCF(-2)
0.026983
0.037754
0.714721
0.4820
GCF(-3)
0.011272
0.037699
0.298990
0.7676
GCF(-4)
-0.043225
0.035015
-1.234466
0.2295
GCF(-5)
0.074039
0.035920
2.061231
0.0508
C
7.482671
1.826004
4.097839
0.0004
R-squared
0.998049
Mean dependent var
26.92435
Adjusted R-squared
0.996777
S.D. dependent var
0.316000
S.E. of regression
0.017940
Akaike info criterion
-4.911122
Sum squared resid
0.007402
Schwarz criterion
-4.228635
Log likelihood
111.7669
Hannan-Quinn criter.
-4.666251
F-statistic
784.4624
Durbin-Watson stat
2.358324
Prob(F-statistic)
0.000000
*Note: p-values and any subsequent tests do not account for model selection.
(Source: Author, 2025)
The ARDL model in Table 4:3 for the manufacturing sector was estimated using 39 adjusted observations from
1985 to 2023, guided by the AIC Criterion which selected the optimal lag structure as ARDL (1, 4, 1, 0, 5). This
model includes one lag of manufacturing output, four lags of renewable energy consumption, one lag of
nonrenewable consumption, no lag of labour and five lags of gross capital formation.
The lag of manufacturing output, MAN(-1), has a coefficient of 0.477181, a standard error of 0.122378, a
tstatistic of 3.899221 and a p-value of 0.0007. This result is statistically significant at the 1% level. It implies
that a 1% increase in manufacturing output in the previous period leads to a 0.477% increase in the current
period, suggesting strong inertia or momentum in the sector. The positive and significant influence of past output
reflects stable production cycles, possibly due to continuous industrial processes, capital stock persistence and
learning effects in Kenya’s manufacturing firms.
Renewable energy consumption has five lags in the model. The contemporaneous value of REC has a coefficient
of 0.071733, a standard error of 0.022235, a t-statistic of 3.226129 and a p-value of 0.0037. This coefficient is
statistically significant at the 1% level and indicates that a 1% increase in renewable energy consumption in the
current year leads to a 0.071% increase in manufacturing output. This confirms that renewable energy sources
such as hydropower, solar and biomass provide an effective, reliable energy input for industrial processes in the
short term.
The first lag, REC (-1), has a coefficient of 0.018555, a standard error of 0.022932, a t-statistic of 0.809136 and
a p-value of 0.4267. This is statistically insignificant, meaning that the immediate past year's renewable energy
consumption does not significantly influence current manufacturing output. Similarly, REC (-2) has a coefficient
of 0.023068, a t-statistic of 0.918899 and a p-value of 0.3677, which is also insignificant. The third lag, REC
(3), has a coefficient of 0.044265, a t-statistic of 1.657385 and a p-value of 0.1110, which remains statistically
insignificant at the 10% level. However, REC (-4), the fourth lag of renewable energy consumption, has a
coefficient of 0.063231, a standard error of 0.025846, a t-statistic of 2.446420 and a p-value of 0.0225. This is
statistically significant at the 5% level, indicating that a 1% increase in renewable energy consumption four years
ago increases current manufacturing output by 0.063%. This lagged effect implies that some forms of renewable
infrastructure or industrial adaptation to clean energy require time to mature and integrate effectively into
production.
Non-renewable energy consumption also appears in two lags. The contemporaneous value has a coefficient of
0.038739, a standard error of 0.063556, a t-statistic of -0.609526 and a p-value of 0.5482, indicating no
statistically significant influence. However, the lagged value, NREC(-1), has a coefficient of -0.113673, standard
error of 0.054764, a t-statistic of -2.075683 and a p-value of 0.0493. This is statistically significant at the 5%
Page 3459
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
level and suggests that a 1% increase in non-renewable energy consumption in the previous year reduces current
manufacturing output by 0.113%. This negative impact may reflect inefficiencies or environmental costs
associated with fossil fuels, including frequent power interruptions, maintenance downtime or emissions-related
regulatory pressure. It also supports the shift toward cleaner energy for industrial use in Kenya.
Labour enters the model without lags and has a coefficient of 0.353834, a standard error of 0.097750, a t-statistic
of 3.619785 and a p-value of 0.0014. This is statistically significant at the 1% level and indicates that a 1%
increase in labour input leads to a 0.353% increase in manufacturing output. The result affirms the
labourintensive nature of Kenyan manufacturing and industrial activities, particularly in small and medium
enterprises where manual operations still dominate. It also indicates high responsiveness of output to workforce
expansion in the short to medium term.
Gross capital formation appears through six terms. The contemporaneous effect has a coefficient of 0.044694, a
standard error of 0.033286, a t-statistic of 1.342713 and a p-value of 0.1925, which is statistically insignificant.
The first four lags, GCF (-1) to GCF (-4), are all statistically insignificant with respective coefficients of
0.031500, 0.026983, 0.011272 and -0.043225 and corresponding p-values above 0.2. However, the fifth lag,
GCF (-5), is statistically significant at the 5% level, with a coefficient of 0.074039, a standard error of 0.035920,
a t-statistic of 2.061231 and a p-value of 0.0508. This indicates that a 1% increase in gross capital formation five
years ago leads to a 0.074% increase in manufacturing output today. This long gestation lag suggests that capital
projects such as plant expansion; machinery investment or infrastructure upgrades may take years to fully impact
industrial sector growth.
The constant term is 7.482671, with standard error of 1.826004, a t-statistic of 4.097839 and a p-value of 0.0004,
which is statistically significant. It represents the base level of manufacturing output when all other variables are
at neutral levels.
The model diagnostic statistics confirm strong performance. The R-squared is 0.998049, indicating that 99.80%
of the variation in manufacturing output is explained by the model. The adjusted R-squared is 0.996777. The
Fstatistic is 784.4624 with a p-value of 0.000000, confirming joint significance. The D-W statistic is 2.358324,
indicating no autocorrelation. The standard error of regression is 0.017940, suggesting very low residual error.
The ARDL results confirm that renewable energy consumption has both immediate and lagged positive effects
on manufacturing output. Labour also contributes strongly, while non-renewable energy exerts a delayed but
negative influence. Gross capital formation has a long-lag positive effect. The findings align with the expectation
that clean, accessible energy and workforce expansion are vital to manufacturing and industrial growth in Kenya,
while fossil energy may increasingly constrain output through operational inefficiencies or rising costs.
ARDL Error Correction Regression
Dependent Variable: D(MAN)
Selected Model: ARDL(1, 4, 1, 0, 5)
Case 3: Unrestricted Constant and No Trend
Date: 07/08/25 Time: 18:14
Sample: 1980 2023
Included observations: 39
ECM Regression
Case 3: Unrestricted Constant and No Trend
Variable
Coefficient
Std. Error
t-Statistic
Prob.
C
7.482671
1.233094
6.068208
0.0000
Page 3460
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
D(REC)
0.071733
0.018564
3.864042
0.0008
D(REC(-1))
-0.130564
0.036179
-3.608797
0.0015
D(REC(-2))
-0.107496
0.029731
-3.615589
0.0015
D(REC(-3))
-0.063231
0.023039
-2.744507
0.0115
D(NREC)
-0.038739
0.048272
-0.802509
0.4305
D(GCF)
0.044694
0.025189
1.774357
0.0892
D(GCF(-1))
-0.069070
0.035721
-1.933606
0.0656
D(GCF(-2))
-0.042086
0.033949
-1.239710
0.2276
D(GCF(-3))
-0.030815
0.028673
-1.074695
0.2937
D(GCF(-4))
-0.074039
0.028380
-2.608823
0.0157
CointEq(-1)*
-0.522819
0.086396
-6.051399
0.0000
R-squared
0.721779
Mean dependent var
0.030105
Adjusted R-squared
0.608429
S.D. dependent var
0.026461
S.E. of regression
0.016558
Akaike info criterion
-5.116250
Sum squared resid
0.007402
Schwarz criterion
-4.604385
Log likelihood
111.7669
Hannan-Quinn criter.
-4.932597
F-statistic
6.367735
Durbin-Watson stat
2.358324
Prob(F-statistic)
0.000044
* p-value incompatible with t-Bounds distribution.
(Source: Author, 2025)
The ECM model in Table 4.4 estimates the short-run dynamics of how renewable and non-renewable energy
consumption affect manufacturing output, while also capturing the sector’s speed of adjustment toward long-run
equilibrium. This ECM is derived from the ARDL (1, 4, 1, 0, 5) model and includes 39 adjusted annual
observations.
The constant term is 7.482671, with a standard error of 1.233094, a t-statistic of 6.068208 and a p-value of
0.0000. This value is statistically significant and reflects the baseline level of manufacturing output in the
absence of shocks from the independent variables.
The change in renewable energy consumption, D(REC), shows a short-run coefficient of 0.071733, a standard
error of 0.018564, a t-statistic of 3.864042 and a p-value of 0.0008. This coefficient is statistically significant at
the 1% level and implies that a 1% increase in renewable energy consumption in the current period leads to a
0.072% increase in manufacturing output. This strong immediate response reflects the ability of renewable
energy inputs to power production lines, run machinery and stabilize energy availability, particularly in regions
with frequent grid interruptions or expensive fossil energy.
The first lag of renewable energy consumption, D(REC(-1), has a coefficient of -0.130564, a standard error of
0.036179, a t-statistic of -3.608797 and a p-value of 0.0015. This is statistically significant at the 1% level and
indicates that a 1% increase in renewable energy consumption in the previous period reduces current
manufacturing output by 0.131%. Similarly, D(REC(-2) is also negative and significant, with a coefficient of
0.107496, a standard error of 0.029731, a t-statistic of -3.615589 and a p-value of 0.0015. These negative lagged
effects suggest that while renewable energy drives immediate gains, excessive or unstable consumption patterns
may reduce efficiency in subsequent periods, possibly due to adjustment costs or the delayed impact of system
inefficiencies. The third lag, D(REC(-3), is also significant, with a coefficient of -0.063231, a standard error of
0.023039, a t-statistic of -2.744507 and a p-value of 0.0115. This further supports the presence of short-run
volatility in how renewable energy affects manufacturing, where initial gains are followed by dampening effects
in later periods.
Turning to non-renewable energy consumption, the contemporaneous change D(NREC) has a coefficient of
Page 3461
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
0.038739, a standard error of 0.048272, a t-statistic of -0.802509 and a p-value of 0.4305. This is statistically
insignificant, indicating that current fluctuations in non-renewable energy consumption do not have a reliable
short-run impact on manufacturing output.
Gross capital formation also displays mixed short-run effects. The contemporaneous change D(GCF) records a
coefficient of 0.044694, a standard error of 0.025189, a t-statistic of 1.774357 and a p-value of 0.0892. This
result is weakly significant at the 10% level, suggesting that new investments can support manufacturing output,
albeit with limited immediate strength. The first lag, D(GCF(-1), has a coefficient of -0.069070, a standard error
of 0.035721, a t-statistic of -1.933606 and a p-value of 0.0656, which is marginally insignificant at the 5% level
but points toward possible delayed negative effects. Subsequent lags D(GCF(-2) and D(GCF(-3) remain
insignificant, with coefficients of -0.042086, a p-value of 0.2276 and a coefficient of -0.030815 with a p-value
of 0.2937, respectively. By the fourth lag, D(GCF(-4) becomes statistically significant, with a coefficient of
0.074039, a standard error of 0.028380, a t-statistic of -2.608823 and a p-value of 0.0157. This negative effect
implies that gross capital formation undertaken four years earlier exerts a drag on current manufacturing output,
possibly due to depreciation, misallocation or inefficiencies in capital use.
The ECM term, CointEq (-1), has a coefficient of -0.522819, a standard error of 0.086396, a t-statistic of
6.051399 and a p-value of 0.0000. This is statistically significant at the 1% level and reflects the speed at which
short-run deviations from the long-run equilibrium are corrected. The coefficient of -0.522819 indicates that
approximately 52.28% of the gap between the current level of manufacturing output and its long-run equilibrium
level is corrected within one year. It means that after a shock, just over half of the adjustment back to long-term
growth happens within a single period. This moderately fast correction speed is indicative of the manufacturing
sector’s structured processes and its ability to respond to energy or capital shocks through systematic
realignment.
Model diagnostics affirm the statistical strength of the ECM. The R-squared is 0.721779 and the adjusted
Rsquared is 0.608429, meaning that about 72% of the short-run variation in manufacturing output is explained
by the included variables. The F-statistic is 6.367735 with a p-value of 0.000044, confirming joint significance
of the regressors. The standard error of regression is 0.016558, while the D-W statistic of 2.358324 suggests no
concern for serial correlation.
The ECM results indicate that renewable energy consumption exerts both immediate positive and delayed
negative effects on manufacturing output, reflecting short-run efficiency gains followed by adjustment costs.
Non-renewable energy consumption has no significant immediate effect, underscoring its diminishing role in
supporting industrial expansion. Gross capital formation shows weak contemporaneous support but reveals
negative lagged effects, pointing to inefficiencies in long-term capital utilization. The moderately fast speed of
adjustment confirms that manufacturing responds systematically to shocks, with steady convergence back to its
long-run growth path, ensuring resilience and stability in the sector.
F-Bounds Test
Null Hypothesis: No levels relationship
Test Statistic
Value
Signif.
I(0)
I(1)
F-statistic
6.238867
10%
2.45
3.52
K
4
5%
2.86
4.01
2.5%
3.25
4.49
1%
3.74
5.06
(Source: Author, 2025)
Page 3462
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
The ARDL bounds test in Table 4:36 yields an F-statistic of 6.238867, which exceeds the upper bound critical
value at the 5% level of 4.01 and even the 1% level of 5.06. This confirms the existence of a long-run equilibrium
relationship between manufacturing output and the regressors, including renewable energy consumption. The
null hypothesis of no cointegration is therefore rejected.
(Source; Author,2025)
As per the results in Figure 4:1, residuals pass the J-B test with a p-value of 0.699138, indicating normal
distribution.
Breusch-Godfrey Serial Correlation LM Test:
F-statistic
1.654110
Prob. F(2,21)
0.2152
Obs*R-squared
5.307694
Prob. Chi-Square(2)
0.0704
(Source: Author, 2025)
The Breusch-Godfrey LM test in table 4:5 yields an F-statistic of 1.654110 and a p-value of 0.2152, which is
statistically insignificant. Thus, there is no evidence of serial correlation in the residuals, confirming that the
model's assumptions are valid.
F-statistic
0.616681
Prob. F (15,23)
0.8319
0
1
2
3
4
5
6
7
8
-0.03
-0.02
-0.01
0.00
0.01
0.02
0.03
0.04
S
e
r
i
e
s
:
R
e
s
i
d
u
a
l
s
S
a
m
p
l
e
1
9
8
5
2
0
2
3
O
b
s
e
r
v
a
t
i
o
n
s
3
9
Mean
1.78e-15
Median
0.000523
Maximum
0.038992
Minimum
-0.028206
Std. Dev.
0.013957
Skewness
0.314716
Kurtosis
3.210512
Jarque-Bera
0.715813
Probability
0.699138
Heteroskedasticity Test: Breusch
-
Pagan
-
Godfrey
Page 3463
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
Obs*R-squared
11.18623
Prob. Chi-Square (15)
0.7393
Scaled explained SS
4.300047
Prob. Chi-Square (15)
0.9966
(Source: Author, 2025)
The Breusch-Pagan-Godfrey test in Table 4:6 shows an F-statistic of 0.616681 and a p-value of 0.8319 and an
Observed R-squared value of 11.18623 with a p-value of 0.7393, both of which are highly insignificant. This
indicates homoskedasticity, a desirable property showing that the variance of errors is constant across
observations.
CUSUM 5% Significance
As per Figure 4:2, the stability plot lies within the 5% critical bounds, indicating structural stability in the model.
Therefore, Manufacturing sector output is positively influenced by renewable energy consumption in both short
and long runs, while non-renewable sources appear to suppress growth. The model is statistically sound, passes
all diagnostic checks and supports the integration of green energy into Kenya’s manufacturing development
policy.
In the manufacturing sector, renewable energy consumption delivers an immediate and statistically significant
positive effect on output, suggesting that renewables can quickly support industrial activities such as machinery
operation and energy stabilization. However, lagged coefficients turn negative, reflecting possible adjustment
costs, inefficiencies in integration or over-reliance on unstable renewable inputs. This volatility highlights the
transitional challenges of embedding renewable energy into industrial production chains. Non-renewable energy
consumption, on the other hand, has no significant short-run impact, pointing to its reduced importance for
immediate manufacturing growth. Gross capital formation provides limited contemporaneous benefits but
eventually generates contractionary effects at longer lags, underscoring inefficiencies in capital absorption and
project alignment. The adjustment speed of just over 52% annually confirms that manufacturing absorbs shocks
at a moderately fast pace, balancing initial energy gains with subsequent corrections.
Therefore, the findings of this study explicitly align with Kenya’s Vision 2030 and SDG 9 by emphasizing
renewable energy as a driver of sustainable industrialization and green manufacturing. Policy interventions
should focus on incentivizing clean energy adoption in industrial zones, supporting technology transfer for
energy-efficient production and also integrating renewable energy infrastructure into manufacturing value
Page 3464
www.rsisinternational.org
ISSN No. 2321-2705 | DOI: 10.51244/IJRSI |Volume XII Issue X October 2025
chains. Linking these strategies to global green growth frameworks will enhance the sector’s competitiveness
while advancing Kenya’s transition toward low-carbon industrial development. This study, therefore, calls for
interventions aimed at increasing renewable energy uptake, for instance, through measures such as removing
taxes on renewable energy consumption and adopting the carbon tax policy so as to increase the uptake of
renewables. This will not only ensure sustainability, but also ensure environmental preservation.
Data aggregation is still a key problem. Total renewable energy consumption is still aggregated hence making
policy intervention to specific renewable energy sources cumbersome. This is because there exists a lot of
renewable energies from sources such as solar that are generated and consumed by firms on-site without the data
being captured anywhere in the national statistics. Efforts should be made by the ministry of energy to streamline
and regulate the ballooning adoption of energy infrastructure such as solar so that their impact on the ground is
correctly captured.
There is need objectively disaggregate renewable energy consumption into their smaller sources for more
informing policy interventions. This will be after serious investments into capturing real data at the firm-level.
1. Bowden, N., & Payne, J. E. (2010). Sectoral analysis of the causal relationship between renewable and non-
renewable energy consumption and real output in the US. Energy Sources, Part B: Economics, Planning,
and Policy, 5(4), 400-408.
2. Das, K. C., & Mahalik, M. K. (2023). Renewable energy use and export performance of manufacturing
firms: Panel evidence from six industries in India. Energy Economics, 125, 106894.
3. Grainger, C. A., & Zhang, F. (2017). The impact of electricity shortages on firm productivity: Evidence from
Pakistan. World Bank Policy Research Working Paper (No. 8130). World Bank.
4. Guliyev., & Yerdelen Tatoglu, F. (2023). The relationship between renewable energy and economic growth
in European countries: Evidence from panel data model with sharp and smooth changes. Renewable Energy
Focus,46, 185-196. https://doi/10.1016/j.ref.2023.05.004
5. Forkuoh, S. K., & Li, Y. (2015). Electricity power insecurity and growth of small manufacturing firms: A
case study of the cold store operators in the Asafo market area of the Kumasi metro in Ghana. Open Journal
of Business and Management, 3(3), 312–322.
6. Hoang, T. H. V., Shahzad, H. J. S., & Czudaj, R. L. (2020). Renewable energy consumption and industrial
production: A disaggregated time-frequency analysis for the U.S. Energy Economics, 85, 104433.
https://doi.org/10.1016/j.eneco.2019.06.018
7. Jeon, H. (2022). Renewable versus non-renewable: The role of electricity generation to economic growth.
The Electricity Journal, 35(6), 107140.
8. Kasae, P. (2014). Energy efficiency and operational performance of manufacturing firms in Kenya. Journals
in Business & Management, 2(3), 15–24.
9. Obange, N., Siringi, E. M., & Mukras, M. S. (2013). Causality between disaggregated energy consumption
and manufacturing growth in Kenya: An empirical approach.
10. Onuonga, S. M., Etyang, M., & Mwabu, G. (2008). The demand for energy in the Kenyan manufacturing
sector. The Journal of Energy and Development, 34(1/2), 265–276.
11. Ototo, E. G., & Nzai, C. (2021). Energy consumption and performance of the manufacturing sector in Kenya.
12. Oyeleke, O. J., & Akinlo, T. (2020). Energy generation and economic growth: empirical evidence from
Nigeria. Environment, Development and Sustainability, 22(8), 7177-7191.
13. World Bank. (2021). World development report 2021: Data for better lives.
14. Zhang, Z., & Ma, Y. (2023). The impact of energy consumption structure transformation on firms’ total
factor productivity: evidence from China. Environmental Science and Pollution Research, 30(31),
7695076968.
