ANALYSIS OF THE GLOBALIZATION OF HUMAN CAPITAL ON THE EXAMPLE OF THE REPUBLIC OF ARMENIA

Yedigaryan Knarik

How to cite this paper: Yedigaryan, K. (2023). Analysis of the Globalization of Human Capital on the Example of the Republic of Armenia. Messenger of ASUE, 1(73), 73-86. DOI: 10.52174/1829-0280_2023.1-73 Received: 15.02.2023. Revision: 16.03.2023. Accepted: 09.06.2023.

KNARIK YEDIGARYAN

PhD Student at the Chair of International Economic Relations at

Armenian State University of Economics

https://orcid.ore/0000-0001-7860-6668

ANALYSIS OF THE GLOBALIZATION OF HUMAN CAPITAL ON THE EXAMPLE OF THE REPUBLIC OF ARMENIA

Human capital is a stock of health, knowledge, skills and abilities formed and accumulated through investments, which is expediently used in the labor process, contributing to an increase in labor productivity and income. As a type of capital, human capital (hereinafter referred to as HC) is quite flexible in terms of globalization. As the banking and financial system plays an important role in the movement offunds, so does the migration of the HC mobility. That is why it is important to understand the relationship between migration and HC.

Developing countries (for example, the Republic of Armenia) are still lagging behind in the creation and effective use of HC, there are not enough conditions for creating a favorable environment for employees providing them with high wages, and ensuring professional growth. The latter is the main reason for the mobility of HC to move from one country to another, in order to use their own HC more efficiently.

The study of the international movement of HC can be useful in the process of HC developing policies, implementing reforms in the areas of education, health care and social protection. The results of the work give an opportunity to develop a migration policy in the interests of the RA economy.

Keywords: human capital, migration, labour market, income, mobility JEL: F22, J24

DOI: 10.52174/1829-0280_2023. 1 -73

Introduction. The role of HC, the expansion of its structure and functions brings a necessity of the formation of a new concept. In comparison with other factors of production that change quantitatively and qualitatively but remain the same in terms of content, HC changes in essence generate new knowledge. In the era of the information revolution, the need to accumulate and develop HC is reflected in the economic policy priorities of the developed countries, which are gradually paying more attention to education and public health.

According to the human capital index calculated by the World Bank, the RA ranks 82nd among 189 countries with an index of 0.6 (World Bank, Human Capital Index, 2020) which means that we are lagging behind many countries in the world on this

indicator. According to the GlobalEconomy.com website, the Brain Drain index 2021, Armenia ranks 42nd among 177 countries with a coefficient of 6.7, which is a fairly high indicator.

From the viewpoint of the globalization of HC, we are mostly interested in the impact of migration on the labor market of the country. The RA is not a leading industrial country, it has no access to the sea, and has an unfavorable geographical position, so the purpose of the given article is to study HC as a factor of internationalization of the RA economy.

Literature review. S. Fan and O. Stark (2013) argue that, theoretically, the migration of HC has a negative impact on the income of the labor force remaining in the country, since labor efficiency decreases. In this case, the employer must make a choice between the unemployed and the employed. L. Shkuflich and V. Vuckovic (2018) show that emigration leads to an increase in unemployment if there are those who leave the country among the employed, and if the gap between the qualifications of the remaining labor force and the outgoing labor force is large. According to B. Elsner (2015) the impact of emigration on income largely depends on the qualifications and quantity of the workforce.

J. Groisard and J. Lull (2006) argue that the future impact of migration of HC, if any, is very small and does not compensate for the losses that the country suffers in short-term capital. In any case, if a country fails to prevent migration, the result is a know-how and an exchange of experience (Mayr, K. and G. Peri, 2009).

Thus, people who study at foreign scientific centers and universities share their experience to a large extent, and the scientific diaspora as a whole has an unprecedented impact on the development of the country. In the case of the RA, the diaspora has repeatedly proven this, and this is not only about the scientific diaspora. Returned migrants often create scientific centers, structures, and engage in entrepreneurial activities.

There are also economic schools that believe that not only migration affects HC, but vice versa, that is, they are interconnected and have a causal relationship. International experience shows that investing in HC can significantly affect the migration behavior of people and vice versa, migration flows, in their turn, affect the HC of a given country.

Some economists believe that higher education in these countries will lead to overtraining and underutilization of specialists. Of course, this does not mean that we should not invest in education, on the contrary, we should train personnel who will subsequently dictate market conditions, we should prepare entrepreneurs, and develop a long-term strategic investment plan in the direction of being a recipient, not a donor of migration. Both qualified and unskilled people are leaving the RA. However, there are no convincing statements about the connection between migration and HC in the economic literature. This issue remains controversial.

Ch. Dustmann and A. Glitz (2011) refer to the work of L. Sjaastad (1962), according to which the decision to migrate cannot be considered in isolation, additional investments in HC are as important as the migration process itself, and sometimes even more important. They argue that the decisions to migrate and to invest in HC are interrelated and that these decisions matter not only for those who leave the country, but also for those who stay there. In their view, it matters who decides to migrate, and the quality of HC in a given country can vary depending on whether or not there is a brain drain or brain gain in that country.

The literature of the last decade argues that skilled workers tend to emigrate to developed countries in the hope of earning higher wages. Typically, a skilled workforce widens the gap between rich and poor countries, making rich countries even richer and poor countries even poorer. Just as in the previous century the flow of financial capital

and the facilitation of its movement greatly changed the international economic environment, it is possible that in the near future this change will be caused by HC.

D. Kapoor and J. McKeil (2005) argue that migration flows are driven by three powerful long-term capitalist trends in attracting migrants: the government policies of rich countries, the deepening globalization of production and trade, which creates great opportunities for the division of labor, and the fact that the integration of capital markets breeds integration of the labor market.

Separate works on HC appeared in Armenian literature only in the 21st century. Our contemporaries G. Gharibyan, Z. Tadevosyan, F. Mayilyan, A. Hovhannisyan, A. Kharatyan, R. Isahakyan, subjecting to a deep analysis of HC, touched upon the issues of its development, reproduction and evaluation in the RA, the features of its implementation.

Perhaps the first who tried to measure HC was W. Patty. He believed that labor was the basis of wealth and that its value should be included in the assessment of national wealth. He tried to calculate HC using the income method, capitalizing wages. L. Dublin, A. Lotka, N. Farr, A. De Foville, A. Barriol, I. Fisher, A. Woods, B. Kiker were also proponents of the income method.

The next method was the cost method created by E. Engel, which was based on the calculation of costs or investments made in HC (E. Engel, K. Dagum, S. Slottje, V. Pareto, Ph. Mahlup):

According to the Jorgenson-Frauman method (Jorgenson, D. and Fraumeni B., 1989), HC is calculated as follows: the total life income (V) of a person of a certain gender, age and education is:

VS,a,e = ys,a,e + Ss,a+1 VS,a+1,e (1 + 9)/(1 + 0 where Y is annual earnings and a S a+1 is the probability that the person will survive another year. The real income growth rate and the discount rate during one year period are denoted respectively by g and i . Jorgenson and Fraumeni identified five stages of the life cycle: no school and no work (ages 0-4), school but no work (5-13), school and work (14-34), work but no school (35-74), and no school or work (75 and older). By assumption, the lifetime income for the oldest group is zero, so is the annual income of those in the first two stages.

We are studying the globalization of HC, so the most acceptable calculation option is the formula proposed by T. Mitze, which includes migration and labor market indicators. T. Mitze (2012) followed the model of J. Harris and M. Todaro (1970), which is based on the differences between the expected incomes of countries of origin and destination and differences in the economic behavior of an individual arising from the employment possibility.

According to this model, the expected income in the country of origin is , and the real wage rate is i(Y0, the probability of finding a job is Pr(Emp,). which can also be considered as a function of labor utilization in country i, other economic and non-economic factors: Xi. Similar designations were made for country designation j. In this case, a person decides to move from country i to country j if:

la < ly ~cij

Where Y^ is the expected income in the country of destination, C, . costs associated with moving from one country to another.

Methodology. In order to describe the entire perspective of the globalization of HC under the influence of migration, we examined the impact of migration on labor market indicators: wages (income), participation in the labor market and labor underutilization, since the accumulation of HC is focused on participation in the labor market, efficient use of HC and income accumulation. In their turn, differences in these indicators are main drivers of HC mobility.

Net migration data have been taken from the UN database (UN Population Division Data Portal, 2022) for all countries. The average wages from the database of the Organization for Economic Cooperation and Development (OECD Data Portal, 2022) and for the RA and the RF from the Federal State Statistics Service of the RF (RG FSSS) and the Statistical Committee of the RA (SC RA), respectively. Labor force participation and labor underutilization rates for all countries have been taken from the International Labor Organization database (ILO).

The World Bank includes the following indicators into the HC index - average years of schooling, life expectancy to age 5, healthy growth, standardized test scores, or adult life expectancy index as the main components of HC. From the point of view of the globalization of the HC, the indicator of education is more important for us, since the subject matter of the given article is the migration of the population of working age, and the latter is mainly based on education. Thus, as an indicator of HC, we choose the average number of years of education calculated by UNESCO (excluding years spent on repeating a year course), or rather the average number of years of education of the population over 25 years old, since migration covers precisely this age group (UNESCO). The data collection of the study was carried out for Armenia as a country of origin, the USA, Russia, France, Germany, Spain, the Netherlands and Portugal, as countries of destination. Most of the countries that are targets for migrants from the RA do not have data on how many Armenians live in this country, and if there were, then, naturally, this would be unreliable data, since many of our compatriots live in other countries on illegal grounds. The SC of the RA has presented the countries which Armenians are leaving for, but this is not the reason for us, because we are interested in labor migration. For example, Georgia is among the top five registered inter-state movements from Armenia, but it is no secret that Armenians go to Georgia not for labor migration, but for the purpose of organizing their rest, so in this list we have chosen such countries, which by their level of development and above-mentioned indicators significantly exceed the RA and naturally attract Armenians in terms of labour migration. The basis for choosing destination countries can also be money transfers sent to the RA, but the CB of the RA generally keeps statistics only for the USA and Russia, where the largest number of transfers come from.The calculations were made for 20102021, since before 2010 and after 2021 some indicators for the RA were not calculated. The equations we used should show the correlation between the indicators of origin and destination countries.

When calculating the relationships and adjustments of labor market indicators and migration of countries i and j Mitze proposed to produce the following equation.

MNETij t = a10 + t£n * (L)MNETij + a12 * (LW^.t-i + «h * + «14 * (QtVij,-, + «15 * (L)HCij t_j + Eij t

where MNET^ t reflects logarithmic differences between net migrations of country i and country j for the given time periud. The indicator variables ij refer to the logarithmic differences of the respective variables for countries i and j, t is the time indicator, L is the value of the lag, W is real wages, LF is labor force participation, LU is labor underutilization, HC is a HC. So our VAR model looks as follows: AMNETtj t = a1Q + an * (^AMNET^ t_l + alz * (L^AW^ x_j + a13

* (L)ALFiy_t_I + au * (_L)ALUij t_, + «15 * (.Q&HC^,., + Eijx

where the subscripts ij indicate the logarithmic differences between countries of origin i and countries of destination j, t is the time index, and L is the lag value.

The remaining equations of the VAR model can be presented in the following

way:

AWiht = «20 + «21 * (L)AMNETijt_i + a22 * (L)AWij,t_1 + a23 * (L)ALFijt_i + a2i * (L)ALUij,t_1 + a2S * (DAHC^ x_i + ^ t AWihl = a3o + a3i * (L)AMNETij,t_i + a^ * (L)AWijt_i + a33 * (L)ALFijt_i + a3i * (L)ALUiht_i + a3S * (L)AHCiht_i + sijf ALFijx = a4o + a4i * (L)AMNET^t_i + au * (L)AWijit_i + at3 * (L)ALFiu_i + au * (L)ALUiu_i + aiS * (L)AHClrt_i + £ti t AHt,t = aso + an * (L)AMNETlh- + aS2 * (L)AW^tA + a53 * (L)ALFlh- + aik * (L)ALUlh- + ass * (L)AHClh- + ^t

It is advisable to calculate this model of equations using a vector autoregression (VAR) model. The VAR model is flexible enough to help draw conclusions about the dynamic changes and interactions of several factors (covariates) over time, which depend both on their time lag and on other factors.

Analysis. In the context of the internationalization of HC, we consider the interaction of migration and HC to be important. However, the situation is greatly influenced by labor market conditions in the countries of origin and destination, especially the level of wages, the level of participation in the labor market and the level of underutilization of the labor market.

Let's look at the levels of the above mentioned indicators in the RA.

Table 1

Indicators of net migration, HC, income, labor force participation and labor underutilization in the RA in 2010-20211

Date Net migration (a thousand people) HC (years) Wage ($, year) Labor force participation (%) Labor underutilization (%)

2010 -31.99 11 3,298 62.5 37.5

2011 0010 -29.21 OQ 71 11.19 1 1 0^ 3,479 "377 62.9 36.1 "5/1 Q

2012 2013 001/1 -28. /3 -28.26 07 11.25 11.26 117/1 3,3 7 7 4,292 63 63 AO 34.9 34.2 "3A 1

2014 2015 oni A -27.76 -27.28 OA QH 11.74 11.57 1 1 4,569 4,308 62 60.5 Q 36.1 36.7 "2/1 ^

2016 2017 om Q -26.80 -26.33 0^ Q/1 11 11.19 1 1 4,358 4,421 A OQ1 59.8 59.2 34.5 34.6

2018 2019 2020 -25.84 -25.36 -12.83 11 11.8 11.33 4,291 4,564 4,652 56.5 57.1 54.3 35.3 35.3 35.3

2021 -12.83 11.39 4,866 54.7 35.3

Net migration in the RA in recent years has always had a negative value, which means that more people emigrate from our country, that is, by its nature, the RA is a country of origin for migrants.

2,000.00 1 cnn nn

1,000.00 Rnn nn

H

0.0 -500.0 n 1 ... 1 1 II II ........1 llllll .. il

u ...... ...... r ^ ^ > > 0 ■ 2011 ■ 2012 ■ 2013 ■ 2014 «2015 ■ 2016 ■ 2017 ■ 2018 «2019 ■ 2020 «2021

■ 2010

Figure 1. Net migration in 2010-20212

1 Table 1 is compiled by the author based on UN Population Division Data Portal, UNESCO, RASC, ILO

databases

2 The chart compiled by the author is based on UN data, United Nations, Data portal, Population Division,

Total-net Migration,

The average number of years of education in the Republic of Armenia for the population over 25 is 11 years on average, which is a fairly good indicator compared to other countries.

Figure 2. Human Capital 2010-2021'

From the point of view of income, the average annual salary in dollars was chosen. In the RA, the average salary for 10 years has increased by only $1,500 and currently stands at $4,866, which is quite small compared to destination countries.

80.000 60.000 40.000 20.000 0 -

II II l,.llllllllllllllll

A >fc :ri>

* * S </ / / *

if

2010 2011 2012 2013 12014 2015 ■ 2016 ■ 2017■ 2018 ■ 2019 ■ 2020 ■ 2021

Figure 3. Average annual wages 2010-20214

The share of the labor force (involvement in the labor market) among the working-age population in the RA is 54.7%, and the indicator of underutilization of the labor force is 35.3%. According to these indicators, we are quite far behind developed countries.

3 The chart was compiled by the author based on UNESCO data, UNESCO, Mean years of schooling, http://data.uis.unesco.org/Index.aspx.

4 The chart was compiled by the author based on data from the UN database, the Statistical Committee of the Republic of Armenia and the Federal Statistical Service of the Russian Federation.

■ 2010 B2011 12012 12013 B2014 B2015 12016 12017 12018 B 2019 1 2020 1 2021 Figure 4. Labour force participation and labour underutilization 2010-20215

Evaluation of a PVAR model requires the satisfaction of several criteria such as stationarity and reliability of the data. To clarify the issue of stationarity of the series more clearly, we performed an Augmented Dickey-Fuller Fisher Unit Root Test for each variable. As a result, it turned out that some of the variables do not satisfy the stationarity condition. To avoid this, the calculation was made on the basis of first-order differentials of the growth rates of indicators, as a result of which all indicators satisfied the stationarity condition.

Summary statistics of indicators are presented in Table 2.

Table 2

Summary statistics

MNET WAGE LU LFP HC 1

Mean 0.023153 0.003624 n nnn/iTO 0.023171 0.000357 0.001875

Median Maximum 0.015668 104.0868 1 no 1 1 AH 0.0094 78 0.616720 i) A i) 111A 0.000000 0.210559 i) ClClpiAGA 0.002635 10.09204 0.004 791 0.198283 n 1C0711

Minimum Std. Dev. 102.214/ 17.58788 i) 1 A A 1Q1 0.403334 0.146560 i) /100^*7/1 0.990654 0.154856 1 7Q11 in 9.179095 1.646129 n oincno 0.158 721 0.067757 n 1 £0 0*7*7

Skewness Kurtosis 0.164383 34.80900 0.4995 74 6.574940 A i) 1 071Q 3. 782110 23.53726 0.829809 34.98659 0.1589 7 7 3.250851

Jarque-Bera Probability Sum 2951.436 0.000000 1.620735 40.18728 0.000000 0.253677 139 7.0 73 0.000000 1.621991 2992.19 7 0.000000 0.025006 0.4 78395 0.787259 0.131275

Sum Sq. Dev. 21344.02 1.482116 1.654644 186.9721 0.316775

Another condition is stability. The inverse roots of the characteristic polynomial carry out the stability. This procedure is extremely important because the non-stationary model is not significant. The model is considered stable if the modules of the inverse roots of the characteristic polynomial are less than 1. Figure 5 represents the values of the inverse roots of the characteristic polynomial by constructing a unit pie chart.

5 The chart was compiled by the author based on data from the International Labor Organization,

Inverse Roots of AR Characteristic Polynomial

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

Figure 5: Graph of the inverse roots of the characteristic polynomial

As we can see, all points are inside the circle, so the values are less than 1. This means that there is no instability in our model.

We consider the structure of the model lag according to several criteria - Akaikey, Schwartz, and Hannan-Quinn criteria.

iНе можете найти то, что вам нужно? Попробуйте сервис подбора литературы.

Table 3

Lag Selection Criteria for Vector Autoregression

12.85906 56.73029 86.66244 116.5014 205.0685

NA 119.2960 44.18556 36.94342 88.56711*

1.61e-06 1.95e-07 1.62e-07 1.48e-07 9.60e-09*

0.850431 1.272871 1.507735 1.738160 4.765165*

1.057297 0.031678 0.767785 1.571687 0.420991*

0.926256 0.817925 0.673667 0.524970 3.172854*

1.5

1.0

0.5

0.0

-0.5

-1.0

-1.5

SC

Table 3 shows that all criteria require the choice of the 4th lag. However, it is necessary to evaluate the presence of autocorrelation in model errors. To do this, we will perform a vector autoregressive residual serial autocorrelation test (VAR Residual Serial Correlation LM tests). As a result, it turns out that for the 1st, 2nd, 3rd lags, the hypothesis of 0 probability is rejected, and for the 4th lag, the probability of 0.74 is confirmed. Therefore, we choose the 4th lag.

We also ran a Granger causality test to see how well the variables explain each other. Under the null hypothesis, variable X does not explain the behavior of variable Y. The test shows that migration and income levels are best explained by labor underutilization, labor underutilization by migration, labor market participation by HC, HC by labor underutilization, and income levels. At general levels, the test shows that the variables we have chosen explain each other.

Table 4

Sequential Autocorrelation Test for Vector Autoregressive Residuals

I Null hypothesis: No serial correlation at lag h 1

Lag | LRE* stat Vdf 1 Prob. Rao F-stat f Prob.

1 67.37282 ZH ClACl? 1 25 0.0000 i) nnm 3.152174 (25, 135.2) 0.0000 pi nnm

2 3 4 5/.96931 55.22312 20.02236 25 25 25 0.0002 0.0005 0.7457 2.620 717 2.471829 0.790857 (25, 135.2) (25, 135.2) (25, 135.2) 0.0002 0.0005 0.7479

Null hypothesis: No .serial correlation at lags 1 to h

Lag I LRE * stat ]df I Prob. I Rao F-stat 1 df I Prob.

1 67.37282 25 0.0000 3.152174 (25, 135.2) 0.0000

2 166.6610 50 0.0000 pi plplplpl 4.877656 (50, 144.7) Hi no o\ 0.0000 pi plplpli)

3 4 217 7680 265.2841 o 100 0.0000 0.0000 4.6954 72 4.788128 ( O, 128.8) ||(100, 107.1) 0.0000 0.0000

Next, we will carry out a detailed analysis of impulses and responses of variables. As a result of such an analysis, it becomes clear what effect one standard deviation of the change in the variables included in the model, impulses, will have on the model. The analysis was carried out using impulse responses and their dispersion decomposition using the Cholesky method. Impulses and responses in standard deviation are shown in Figure 6, the decomposition of the variance is shown in Figure 6 as well. To understand the interaction of variables in short, medium and long-term capitalizations, we discussed them in years 1, 5 and 10. Figure 6 illustrates the abovementioned.

Table 5

Granger Causality Test

Dependent variable: MNET

Excluded Chi-sq df Prob.

WAGE 1.893041 4 0.7554

LU 9.746257 4 0.0449

LFP 2.459471 4 0.6519

HC 4.402040 4 0.3543

All 15.15005 16 0.5137

Dependent variable: WAGE

Excluded Chi-sq df Prob.

MNET 0.751176 4 0.9449

LU 20.46579 4 0.0004

LFP 2.695106 4 0.6101

HC 5.797847 4 0.2148

All 40.10568 16 0.0008

Dependent variable: LU

Excluded Chi-sq df Prob.

MNET 83.69932 4 0.0000

WAGE 1.253835 4 0.8692

LFP 1.226620 4 0.8737

HC 6.469064 4 0.1668

All 104.0686 16 0.0000

Dependent variable: LFP

Excluded Chi-sq df Prob.

MNET 4.991365 4 0.2882

WAGE 2.810186 4 0.5901

LU 6.395651 4 0.1715

HC 11.74739 4 0.0193

All 92.77622 16 0.0000

Dependent variable: HC

Excluded Chi-sq df Prob.

MNET 0.495204 4 0.9740

WAGE 13.16432 4 0.0105

LU 15.33180 4 0.0041

LFP 1.576132 4 0.8131

All 57.13773 16 0.0000

Response to Cholesky One S.D. (d.f. adjusted) Innovations ± 2 S.E.

Response of DMNET to DMNET

Response of DM NET to DWAG E

Response of DMNET to DLU

Response of DM NET to DLFP

/\

* _____________-V- _______________________

V-.....V.A- * ----------------\v/ _ s/

•40 •40 ■ 40 ■4°-

Response of DMNET to DHC

Response of DWAGE to DM NET

Response of DWAGE to DWAG E

Response of DWAGE to DLU

Response of DWAGE to DLFP

Response of DWAGE to DHC

Response of DLU to DHC

2 4 6 8 10

Response of DLFP to DMNET

2 4 6 8 10

Response of DLFP to DWAGE

Response of DLFP to DLU

Response of DLFP to DLFP

Response of DLFP to DHC

_______ 04 04 ^ A >' 04

-.4 .......''v'C,I. -.4 v ^VAv^N -04 ,,t..... V

Response of DHC to DMNET

Response of DHC to DWAGE

Response of DHC to DLU

Fgure 6. Answers in standard deviation

Response of DHC to DLFP

Response of DHC to DHC

Table 6

Cholesky method of decomposition of responses by standard deviation

iНе можете найти то, что вам нужно? Попробуйте сервис подбора литературы.

Variance Decomposition of DMNET: \