Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
21
ITALIYADA IQTISODIY RIVOJLANISH VA TANLANGAN IQTISODIY
KO'RSATKICHLAR O'RTASIDAGI BOG'LIQLIK
Amirov Xurshid Axrol o’g’li
Toshkent davlat iqtisodiyot universiteti,
Bafoyev Azizbek Vohidovich
Toshkent davlat iqtisodiyot universiteti,
Durdimurodov Bahodir Botirovich
Toshkent davlat iqtisodiyot universiteti,
Zayniddinov Rukhiddin Xusniddin o'g'li
Toshkent davlat iqtisodiyot universiteti
Annotatsiya.
Bu yerda biz Italiyaning iqtisodiy rivojlanishini o'lchaymiz va unga ta'sir qiluvchi
omillarni o'rganamiz. Iqtisodiy o'sishga katta ta'sir ko'rsatishini o'rganish uchun biz bir nechta omillarni
tanlaymiz. Biz aholi jon boshiga YaIMni mamlakatning iqtisodiy rivojlanishining o'lchovi sifatida olamiz.
Ushbu tadqiqot o'zgaruvchilar va aholi jon boshiga to'g'ri keladigan YaIM o'rtasidagi dinamik va uzoq
muddatli munosabatlarni o'rganishga qaratilgan.
Kalit so‘zlar:
aholi jon boshiga yalpi ichki mahsulot, ko'p o'zgaruvchan vaqt seriyasi, VAR modeli,
valyuta kursi, ishsizlik, eksport, sanoat.
ВЗАИМОСВЯЗЬ МЕЖДУ ЭКОНОМИЧЕСКИМ РАЗВИТИЕМ И ОТДЕЛЬНЫМИ
ЭКОНОМИЧЕСКИМИ ПОКАЗАТЕЛЯМИ НА ПРИМЕРЕ ИТАЛИИ
Амиров Хуршид Ахрол угли
Ташкентский государственный экономический университет,
Бафоев Азизбек Вохидович
Ташкентский государственный экономический университет,
Дурдимуродов Баходир Ботирович
Ташкентский государственный экономический университет,
Зайниддинов Рухиддин Хуснидин угли
Ташкентский государственный экономический университет
Аннотация.
В данной статье мы измеряем экономическое развитие Италии и изучаем
факторы, которые на него повлияют. Для изучения мы выбрали несколько факторов, которые
оказывают большое влияние на экономический рост. Мы принимаем ВВП на душу населения в
качестве меры экономического развития страны. Целью данного исследования является
изучение динамических и долгосрочных связей между переменными и ВВП на душу населения.
Ключевые слова:
ВВП на душу населения, многомерные временные ряды, модель ВАР,
обменный курс, безработица, экспорт, промышленность.
VII SON - NOYABR, 2023
UO‘K: 330.341
21-31
Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
22
THE RELATIONSHIP BETWEEN ECONOMIC DEVELOPMENT AND SELECTED
ECONOMIC INDICATORS IN CASE OF ITALY
Amirov Xurshid Axrol ugli
Tashkent state university of economics,
Bafoyev Azizbek Vohidovich
Tashkent state university of economics,
Durdimurodov Bahodir Botirovich
Tashkent state university of economics,
Zayniddinov Rukhiddin Khusnidin ugli
Tashkent state university of economics
Abstract.
In this article, we measure the economic development of Italy and examine the factors that
will influence it. For the research purposes we have selected several factors that have a great impact on
economic growth. We take GDP per capita as a measure of a country's economic development. The purpose
of this study is to examine the dynamic and long-term relationships between variables and GDP per capita.
Keywords:
GDP per capita, multivariate time-series, VAR model, exchange rate, unemployment,
export, industry.
Introduction.
Studying economic development of countries is important for several reasons. Understanding the
components that lead to economic growth: By studying economic development, we can learn more
about the elements that support economic growth, such as industry, investment, and innovation, as well
as the elements that discourage it, such as inflation and unemployment. Identifying opportunities for
trade and investment: Economic development studies can help identify countries and regions that offer
opportunities for trade and investment. And also understanding global economic trends: Economic
development studies can help us understand global economic trends and how they affect different
countries and regions. As there are many cases of global trends, downturns, recessions, whose causes
and effects to economic development are clear. This information can be used to construct policies that
support economic growth and development by economists and policymakers. So, we opted to check and
find the answer for the question of whether link between independent and dependent variables exist or
not.
Literature review.
Regarding these variables, there are several scientists that studied link between variables we
chosen and economic growth. To begin with Simon Kuznets (1946) book of National Income, which says
GDP per capita would be best variable to show one nation’s well
-being, its welfare and thus urged us to
take GDP per capita as dependent variable of in research.
Additionally, exchange rate, in Paul Krugman (1986) states: exchange rate fluctuations could have
a significant impact on economic growth in his paper called "Target Zones and Exchange Rate
Dynamics", indicating there is a link between them.
For unemployment, one of the scientists that studied it is Arthur Okun (1962), which is famous
for his Okun law. In his book called
"
Potential GNP: Its Measurement and Significance
"
he states that 1
% increase in unemployment leads to 2 % decrease in GDP.
As for industry, Robert Solow (1957) finding is worth to look at. In his paper "Technical Change
and the Aggregate Production Function,", he argued that Economic expansion was mostly fueled by
technological advancement, and that industry was essential to this process.
However, there are some scholars who argues about the findings of above mentioned scientists.
As for Paul Krugman’s (1986) statement, Theodore McKinnon Exchange rate stability, according to
McKinnon's (1993) essay "The Rules of the Game: International Money and Exchange Rates," was crucial
for economic expansion.
For unemployment Robert Lucas (1976): Lucas argued that other factors, such as productivity
growth, were more crucial for economic growth in his 1976 paper "Econometric Policy Evaluation: A
Critique," which challenged Okun's law, which suggests a negative relationship between unemployment
and economic growth.
Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
23
As a result, there are different opinions about the link between indicators and economic growth.
Therefore we decided to look at this in case of Italy between the period of 1970-2021 to see which is
true for Italy. Data for these variables are taken from official site of FRED for the period given above
Methodology.
We have opted for a quantitative method utilizing a multi-factor time-series model to determine
the link between GDP per capita and various parameters.
The variables chosen for this hypothesis test were as follows:
-exchange rate, unemployment rate, export, industry were selected as independent variables
-GDP per capita, as a measure of economic development, was selected as a dependent variable.
Figure 1. Model
Our hypothesis as follows:
H0: There is no link between exchange rate and GDP per capita.
H1: There is a link between exchange rate and GDP per capita.
H0: There is no relationship between unemployment rate and GDP per capita.
H2: There is relationship between unemployment rate and GDP per capita.
H0: There is no connection between export and GDP per capita.
H3: There is a connection between export and GDP per capita.
H0: There is no link between industry and GDP per capita.
H4: There is a link between industry and GDP per capita.
6
created by author
Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
24
Figure 2. Hypothesis testing
In particular, we created an econometric model and equations utilizing multi-factor time series,
examining selected factors and the per capita GDP value for the years 1970
–
2021.
The following model is developed to investigate the link between variables and GDP per capita:
ln
𝐺𝐷𝑃𝑝𝑒𝑟𝑐𝑎𝑝𝑖𝑡𝑎
l =
𝛽
0
+
𝛽
1
lnexhangerate
𝑖
+
𝛽
2
lnunemployment
𝑖
+
𝛽
3
lnexport
𝑖
+
𝛽
4
lnindustry
𝑖
+
𝜀
𝑖
lnGDPpercapita: natural logarithm of GDP per capita
lnexhangerate: natural logarithm of exchangerate
lnunemployment: natural logarithm of unemployment rate
lnexport: natural logarithm of export
lnindustry: natural logarithm of industry
𝛽
0
: the intercept of the model
𝜀
𝑖
: error term
The VAR model specification is given as follows:
𝑌
𝑡
=
𝑎
+
𝛽
1
𝑌
𝑡−
1
+
𝛽
2
𝑌
𝑡−
2
+
⋯
+
𝛽
𝑝
𝑌
𝑡−𝑝
+
𝜀
𝑖
where α is the intercept, a constant and β1, β2 till βp are the coefficients of the lags of Y till order p.
Order ‘p’ means, up to p
-lags of Y is used and they are the predictors in the equation.
The ε_{t} is the error, which is considered as white noise.
By utilizing models like VAR in multi-factor time series, we also created a forecast for a few chosen
indicators in our study. We employed the Stata 17 program, which is now popular among scholars all
over the world, in order to model and forecast.
In multi-factor time series, the cointegration relationship was performed in the following steps:
-indicators were logged;
-time series were checked for stationary;
-a regression model was built;
-the residue was checked for stationary.
Stationary Test
With the Augmented Dickey-Fuller (ADF) test, a unit root is examined. Do the observed variables
typically resume their long-term trend after a shock, or do they randomly wander? The regression
between variables is false if, after a transient or persistent shock, the variables behave randomly.
Therefore, the parameter estimates from the OLS will not be consistent. Every series ought to be level
and stationary. Equation can be used to determine the ADF test.
7
created by author
Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
25
𝑌
𝑡
=
𝛽
1
+
𝛽
2
𝑡
+
δ𝑌
𝑡−
1
+
𝑎
𝑖
∑
𝑚
𝑖
𝑌
𝑡−
1
+
𝜀
𝑡
The hypothesis tested:
H0: δ = 0 (contains a unit root, the data are not stationary)
H1: δ < 0 (does not contains a unit root, the data are stationary).
Before performing the model's prognosis, the direction and density of the indicators were
determined using the six conditions of Gaus Markov, as well as the heteroskedastic problem, the model
residual autocorrelation problem, and regression models.
Analyses and Results.
As mentioned above, the first figure (GDP per capita) is our dependent variable, and the rest of
the figures ( exchange rate, unemployment rate, export, industry) are our independent variables. We
can see that economy of the Italy developed during this period, as we took GDP per capita as its measure
and it had almost reached 35600 USD. And we have conducted some work on which factors have
significant from above influence to it and whether there is positive or negative correlation between
them. Since our study uses multi-factor time series, the first step in the multi-factor time series criterion
is to look at the variables that the Dickey-Fuller test determines to be stationary or non-stationary. Then,
we can choose a certain suited model.
Table 1.
Results of the Dickey-Fuller test on GDP per capita, exchange rate,
unemployment, export and industry respectively
Variable: GDP per capita
Test statistics
1 % critical
value
5 % critical
value
10 % critical
value
Observation
P-value
-6.021
-3.580
-2.930
-2.600
50
0.0000
Variable: exchange rate
Test statistics
1 % critical
value
5 % critical
value
10 % critical
value
Observation
P-value
-5.246
-3.580
-2.930
-2.600
50
0
Variable: unemployment
Test statistics
1 % critical
value
5 % critical
value
10 % critical
value
Observation
P-value
-5.160
-3.580
-2.930
-2.600
50
0
Variable: export
Test statistics
1 % critical
value
5 % critical
value
10 % critical
value
Observation
P-value
-6.893
-3.580
-2.930
-2.600
50
0
Variable: industry
Test statistics
1 % critical
value
5 % critical
value
10 % critical
value
Observation
P-value
-7.064
-3.580
-2.930
-2.600
50
0
Since none of the variables were chosen to be stationary, as can be seen from the example above,
the values of all variables became stationary after one integration
The development of a model of regression and correlation of the impact of chosen economic
indicators assets on the GDP per capita is the next step in achieving the main objective of our study.
Figure 3. Correlation table between variables
8
created by author
9
created by author
industry 0.6663 0.7469 0.4911 0.5175 1.0000
export 0.9603 0.5442 0.3724 1.0000
unemployment 0.4115 0.6972 1.0000
exchangerate 0.5876 1.0000
gdppercapita 1.0000
gdpper~a exchan~e unempl~t export industry
Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
26
It is the result of correlation test. We can see that factors have significant influences with the
biggest is export ( 96 %), which is also the one that has positive correlation, whereas the lowest
unemployment (41 %). Figures for exchange rate and industry negative 58% and 66% respectively. We
can also see that there is a strong correlation between industry and exchange rate which is negative
74%, whereas export and unemployment the least correlated one with negative 37%. Correlation
between other variables has a range of between 49% and 69%. It is also worth mention that all variables
are positively correlated with each other in our case.
We convert the indicators to a natural logarithm and put them in the form of a simple regression
and correlation econometric formula because they were different in the development of a simple
regression and correlation econometric model.
ln
𝐺𝐷𝑃𝑝𝑒𝑟𝑐𝑎𝑝𝑖𝑡𝑎
l =
𝛽
0
+
𝛽
1
lnexchangerate
𝑖
+
𝛽
2
lnunemployment
𝑖
+
𝛽
3
lnexport
𝑖
+
𝛽
4
lnindustry
𝑖
+
𝜀
𝑖
The "Ordinary least squares method" has been used to generate an economic model from the
simple regression and correlation.
The results of this simple regression and correlation econometric model analysis are presented
below
Figure 4. Regression table
The calculations described above are used to create the following four-factor regression model:
ln
𝐺𝐷𝑃𝑝𝑒𝑟𝑐𝑎𝑝𝑖𝑡𝑎𝑖
= -11.83 -0.24lnexchangerate
𝑖
+0.20lnunemployment
𝑖
+ 0.727lnexport
𝑖
+0.86lnindustry
𝑖
+
𝜀
𝑖
Generally, each factor we chosen have significant influence on our dependent variable (GDP per
capita in our case), as all independent variables’ p value is smaller than 0.05. The developed regression
model's Fisher F-statistic has a P-value probability of less than 0.05, showing that the constant and
independent variable component influences GDP per capita. We have 52 observations. R-squared and
adjusted R-squared are also 98% both, meaning factors we chose can explain 98% of changes in GDP
per capita. Other part is explained by some other variables or error term. As for coefficients, exchange
rate’s coefficient is negative 0.24 meaning that 1 unit change inflation leads to 0.24 unit change in GDP
per capita. As for unemployment and industry, figures are 0.20 and 0.86 respectively, showing 1 percent
change in them leads to 0.20 and 0.86 percent increase in GDP per capita. As there should be some
unemployment too for economic growth, as otherwise there would be inflation. So this is why
unemployment has also
positive effect on GDP per capita. Export’s figure is also 0.72, means 1 unit
change in export has effect of 0.47 in GDP per capita. In fact, export has only positive influence on GDP.
Our intercept is negative 11.83.
We continue a diagnostic analysis on this model using globally recognized Gauss-Markov criteria.
According to Gauss Markov's first condition, there should be six times as many observations as
indicators. With 31 observations and 5 indicators, we can see that our model has met the first
requirement of the Gauss-Markov equation.
The empirical model, which is expressed as follows in the table, is equal to the total of the
theoretical data in accordance with Gauss Markov's second condition.
10
created by author
_cons -11.83763 .8363577 -14.15 0.000 -13.52017 -10.1551
lnIndustry .8648694 .1684643 5.13 0.000 .525963 1.203776
lnExport .7258104 .0257383 28.20 0.000 .6740316 .7775892
lnUnemployment .2092923 .0943164 2.22 0.031 .0195522 .3990323
lnexchangerate -.2420766 .097254 -2.49 0.016 -.4377265 -.0464267
lnGDPpercapita Coefficient Std. err. t P>|t| [95% conf. interval]
Total 38.2374959 51 .749754822 Root MSE = .1081
Adj R-squared = 0.9844
Residual .549263908 47 .011686466 R-squared = 0.9856
Model 37.688232 4 9.42205801 Prob > F = 0.0000
F(4, 47) = 806.24
Source SS df MS Number of obs = 52
Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
27
Table 2.
Gaus Markov's 2nd condition on the model
Var
Obs
Mean
Std.dev.
Min
Max
lnGDPpercapita
52
9.65441
0.8658838
7.652956
10.61998
model
52
9.65441
0.8596423
7.548821
10.62664
As there is no negative data which we chose, the number of observations of gdp per capita and our
model is the same. Based on the information in Table, it can be concluded that our model met
requirement 2 with success.
The residue need not be connected to the model, which is the third requirement. A heteroskedastic
state is one in which the residuals and the model are connected. You may check this in three distinct
ways. We'll use the White test and the Breusch-Pagan test for the tests.
We will use the Breusch-Pagan test to start evaluating our model.
Table 3
Breusch-Pagan test result
Chi2(1)
Prob>chi2
lngdppercapita
0.10
0.7576
According to the Breusch-Pagan test results, the test's p value is greater than 0.05, which is
referred to as the homoscedastic state by this test criterion. Additionally, the White test is where we will
test our model next. The p value for this test must be more than 0.05, just like it was for the Breusch-
Pagan test mentioned above.
Figure 5. White test
From the picture above, the White test's p value is higher than 0.05, which disqualifies the
heteroskedastic state in accordance with its criteria and enables us to accept alternative hypothesis 1.
The value of the Shapiro-Wilk test was 0.08 in accordance with Gaus Markov's fourth requirement,
and given that this value is likewise bigger than p 0.05, we can see that this condition is also satisfied.
Figure 6. Shapiro-Wilk test
Figure below shows that the residuals are regularly distributed, with some exception on the right
sight.
11
created by author
12
created by author
13
created by author
14
created by author
Total 22.89 19 0.2424
Kurtosis 0.03 1 0.8536
Skewness 7.12 4 0.1299
Heteroskedasticity 15.74 14 0.3298
Source chi2 df p
qoldiq 52 0.96058 1.912 1.386 0.08290
Variable Obs W V z Prob>z
Shapiro
–
Wilk W test for normal data
Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
28
Figure 7. Histogram. Normal distribution of residuals
Below is test results of checking distribution of residuals. According to sktest our residuals was
perfectly distributed as probability is greater than 0.05 (almost 5 which is also acceptable in case of r)
Figure 8. Sktest of residuals
The 5
th
of Gauss Markov conditions is checking the existence of correlation between independent
variable and there should be no correlation between them. We will use VIF test to check this
Figure 9. VIF results between variables
According to the picture above, there is no any strong correlation between independent variables
as VIF test should be smaller than 10.
The absence of an autocorrelation issue in the model residuals is the sixth need for model
verification. The sixth criterion can be verified in three different methods, using the graph,
autocorrelation table, Durbin-Watson test, and Breusch-Godfrey test.
We will start testing the model using the Durbin-Watson test in the test procedure. The Durbin-
Watson test value falls between 0 and 4 according to the requirements of this test. There is no
autocorrelation if the test result for the model is close to 2. There is an autocorrelation if the result is
between 0 and 1.5 or greater than 2. The outcome of our model's execution using this test was
0.5679359, showing there is some autocorrelation between residuals. Next, we'll use the Breusch-
Godfrey test to see if there are any autocorrelation issues in the residuals.
Figure 10. Breusch-Godfrey test
15
created by author
16
created by author
17
created by author
18
created by author
qoldiq 52 0.0449 0.5852 4.41 0.1104
Variable Obs Pr(skewness) Pr(kurtosis) Adj chi2(2) Prob>chi2
Joint test
Skewness and kurtosis tests for normality
Mean VIF 3.60
lnUnemploy~t 2.44 0.410542
lnExport 3.33 0.300466
lnIndustry 3.43 0.291414
lnexchange~e 5.21 0.192120
Variable VIF 1/VIF
H0: no serial correlation
1 26.190 1 0.0000
lags(
p
) chi2 df Prob > chi2
Breusch
–
Godfrey LM test for autocorrelation
Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
29
We can infer that there is autocorrelation between the residuals from the Breusch-Godfrey test
results. As a result of the R-square probability level being less than 0.05.
These tests showed that our model passed all 6 conditions of the Gauss-Markov test, with the
exception of the last. But we are allowed to go on to the forecasting phase of our research after the
identification and assessment phases.
Firstly, we used VAR model to forecast and see which variables has significant influence on our
dependent variable. If p value is smaller than 0.05, we can say that it has significant influence on GDP
per capita. First, we look at GDP per capita itself, to see whether it has influence. According to GDP per
capita it has no significant influence on upcoming years to itself, as p value is not smaller than 0.05
(which are 0.42 and 0.06).
Figure 11. VAR model of our analysis
Industry and exchange rate posses great influence for both years. P values are 0.035 and 0.056 for
the former, 0.054 and 0.001 for the latter. Industry has 155 % and negative 144 % impact on GDP for
upcoming 2 years. As for exchange rate, 1
st
year it impacts almost 137 times negatively and 2
nd
year 220
times. Other variables or variables in other years do not hold significant influence. The figures for log
likelihood should be as big as possible but negative in our case. This criteria is true for Det(sigma),
AIC,HQIC and SBIC figures and these are 1.83e , 67, 68 and 69 respectively which meets criteria.
The VAR model specification is given as follows:
𝑌
𝑡
= 𝑎 + 𝛽
1
𝑌
𝑡−1
+ 𝛽
2
𝑌
𝑡−2
+ ⋯ + 𝛽
𝑝
𝑌
𝑡−𝑝
+ 𝜀
𝑖
where
α
is the intercept, a constant and
β
1,
β
2 till
β
p are the coefficients of the lags of Y till
order p.Order ‘p’ means, up to p
-lags of Y is used and they are the predictors in the equation. The
ε
_{t} is the error, which is considered as white noise.
19
created by author
_cons 712.7325 2814.034 0.25 0.800 -4802.674 6228.139
L2. 22009.71 6658.023 3.31 0.001 8960.223 35059.19
L1. -13698.13 7103.099 -1.93 0.054 -27619.95 223.692
exchangerate
L2. -144.0712 75.45605 -1.91 0.056 -291.9623 3.81995
L1. 155.9842 74.17125 2.10 0.035 10.61117 301.3571
industry
L2. -7.93e-09 2.02e-07 -0.04 0.969 -4.05e-07 3.89e-07
L1. -1.71e-08 2.06e-07 -0.08 0.934 -4.21e-07 3.87e-07
export
L2. -58.90023 360.0653 -0.16 0.870 -764.6153 646.8148
L1. -495.9692 351.9647 -1.41 0.159 -1185.807 193.869
unemployment
L2. .6363403 .3458271 1.84 0.066 -.0414684 1.314149
L1. .3078189 .3828238 0.80 0.421 -.442502 1.05814
gdppercapita
gdppercapita
Coefficient Std. err. z P>|z| [95% conf. interval]
exchangerate 11 .0628 0.9223 419.0224 0.0000
industry 11 4.44848 0.9296 516.5383 0.0000
export 11 2.8e+09 0.9739 1866.498 0.0000
unemployment 11 .709979 0.9033 398.1079 0.0000
gdppercapita 11 1874.5 0.9804 2504.15 0.0000
Equation Parms RMSE R-sq chi2 P>chi2
Det(Sigma_ml) = 1.83e+22 SBIC = 69.75377
FPE = 1.71e+23 HQIC = 68.45147
Log likelihood = -1636.264 AIC = 67.65055
Sample: 1972 thru 2021 Number of obs = 50
Vector autoregression
Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
30
Y
t
=712.73+156
L1
industry
t-1
-144
L2
industry
t-2
-13698
L1
exchangerate
t-1
+22009
L2
exchangerate
t-
2
+
ε
t
On the basis of the findings from our VAR model, we will forecast GDP per capita for the years
2022 through 2026 in the following phase.
Our study's prognosis was improved by using the VAR that was more successful at doing so, and
as a result, we met our research objective.
Figure 12. Forecasting GDP per capita for the period 2022-2026 according to 3 probability
Figure13. Table of forecast for variables for the period of 2022-2026
It is clear from the table and figure that the dependent variable's forecast ranges from 2022 to
2026.
Additionally, according to projections based on economic metrics we chose, Italy’s GDP per capita
will increase to almost $35000 by 2026. But this figure can reach around 41500 if economy do well and
take into account external factors or can decrease to up to 28500 if there is any kind of sudden condition
like quarantine and etc.
The main objective of the article was to draw attention to how economic variables affected the
nation's economic growth from 1970 to 2021. Therefore, in order to specifically show how the selected
economic indicators can affect the economic advancement in the instance of Italy, we employ the World
Bank's approach to calculate the level of development of countries.
In order to test our hypothesis, we employed a multi-factorial time series to look at the
relationship between GDP per capita growth and both short- and long-term economic indices. 6 Gauss
Markov conditions were used to assess our findings, and our models passed almost all six evaluation
tests (although 6
th
condition was unsuccessful).
Furthermore, we decided to use the VAR model only after our models had passed the identification
and evaluation tests. because when it came to forecasting, the VAR model provided us with useful
results. The research utilizing the multi-factor time series model showed that the impact of economic
indicators on GDP per capita is significant in the case of Italy, which has one of the best economies in the
world. As a result, we can draw the conclusion that both long- and short-term economic growth are
significantly influenced by economic indicators.
20
created by author in stata
21
created by author in stata
Iqtisodiy taraqqiyot va tahlil, 2023-yil, noyabr
www.e-itt.uz
31
There are many famous economists and organizations that support the idea that GDP per capita
is a good indicator of economic development of a country. Such as The World Bank, Nobel laureate
economist Simon Kuznets who is known for developing the concept of GDP and for his work on economic
growth, The International Monetary Fund (IMF) and others. Depending on them we chose GDP per
capita to see economic development. And we would like to know exactly which macroeconomic
indicators have significant impact on GDP per capita, in case of Italy. So we made hypothesis whether
several indicators and our dependent variable has a link. We used a multi-factorial time series to
evaluate how long-term and short-term macroeconomic indicators can effect GDP per capita growth,
primarily utilizing the log-log model (as our variables has different such as $ and % units) and VAR
model ( which is one of the strong models for forecasting). Also We examined our study under 6 different
Gauss Markov settings to see how accurate it was, and all of our models passed five evaluation tests.
Additionally, the stationary status of our factor indicators and residuals was checked; after 1
integrations, all variables became stationary, enabling the use of the VAR model. Through VAR model,
we detected that 2 variables (industry and exchange rate) have significant impact on our dependent
variable for next 2 lags. It also allowed us to forecast and see figures for all variables for next 5 years.
Overall, we can say that all variables almost stay the same or will not change very much except industry
which may face a decrease of 7% until 2026. As for correlation, the most correlated variable with
dependent variable was export(96%) and industry and exchange rate also strongly correlated with
almost 75%. Strangely, all chosen variables have positive correlation with GDP per capita. For
regression, all variables have significantly important with p value smaller than 0.05. Industry has biggest
impact of 0.86% for every 1 percent change. Only exchange rate have negative impact with 0.24 unit for
every 1 unit change. The chosen variables can explain 98% changes of our GDP per capita, which was
known through our R-squared.
Conclusion.
We can conclude that 4 variables, in our case exchange rate, unemployment, export and industry
are proven to have significant influence on economic growth (which is shown through GDP per capita
in our case) . At first we hypothesized whether there is a link between these variables and we proved it.
To do this we used time series and VAR model. We also used D.Fuller test to make our variables
stationary. From regression, we saw 3 our independent variables, except for exchange rate, have
positive impact on GDP per capita. This is always the case for export and industry as they always serve
to increase GDP figure. But for unemployment this can increase GDP up to certain point, and when it
exce
eds it’s norm it will start to impact negatively. As it is supported by some scholars, one of them is
Robert Lucas VAR model helped us to determine for upcoming 2 lags which variables have great impact
on GDP per capita by stating how much effect will result for every percent change. Also we can see 3
possible outcomes of our GDP per capita which all have the same probability to occur according to
situation in world and other external and internal factors.
Literature:
Arthur Okun (1962) Potential GNP: Its Measurement and Significance. Cowles Foundation for
Research in Economics at Yale University.
Paul Krugman (1986) Target Zones and Exchange Rate Dynamics. The Quarterly Journal of
Economics, Oxford University Press.
Robert Lucas (1976) Econometric Policy Evaluation: A Critique. Chicago university press.
Robert Solow (1957) Technical Change and the Aggregate Production Function. Harvard university
press.
Ronald McKinnon (1993) The Rules of the Game: International Money and Exchange Rates. MIT
Press.
Simon Kuznets (1946) National Income: A Summary of Findings. Journal of “Survey of current
business.
