The Impact of Internet Development on International Trade in Services

Number of mobile cellular subscriptions per 100 at origin 0.997 (0.949)

Number of mobile cellular subscriptions per 100 at destination 0.788 (0.811)

Education industry X Mobile subscriptions at origin 0.637*** (0.227)

Construction industry X Mobile subscriptions at origin 0.237 (0.226)

Telecom industry X Mobile subscriptions at origin 0.111 (0.226)

Insurance industry X Mobile subscriptions at origin 0.040 (0.226)

Financial industry X Mobile subscriptions at origin -0.016 (0.226)

Health industry X Mobile subscriptions at origin 0.651*** (0.227)

Constant -530.983*** (50.018)

-------------------------------------------------------------------------------

Observations 2,655

R2 0.391

Adjusted R2 0.379

Residual Std. Error 2.936 (df = 2604)

F Statistic 33.394*** (df = 50; 2604)

===============================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

In model 7 the interaction effects between industry type and number of mobile cellular subscriptions per 100 people at origin are estimated. Again, the results are qualitatively the same as previous models.

As a preliminary result, industry differentiation turned out to be quite robust to model specification.

The `export from developing to developed' trade pattern

The special trade patterns between developing and developed countries are often reported in literature (see Clarke & Wallsten, 2006 for example) (in particular, export from developing countries to the developed ones). To take a closer look at such possible patterns the special type of interaction effects were constructed. The database `ICT Access and Usage by businesses' was used to obtain some indicators of destination which could be used as the proxy-variables for the attractiveness of the particular destination as the export of services via the Internet target. Then these `enterprise variables' were interacted with `Origin is not developed' dummy (the inverse of `Origin is developed').

Table 10. Model 8 estimation results.

===============================================================================

Dependent variable:

-----------------------

Trade flow

-------------------------------------------------------------------------------

Predicted Probability of trade 0.149*** (0.022)

Distance -0.098 (0.077)

Common official language 1.731*** (0.426)

Contiguity -0.849*** (0.291)

Mass variable 5.078*** (0.466)

Origin is developed 16.042*** (2.133)

Number of secure Internet servers per 1M at origin -11.182*** (1.078)

Number of secure Internet servers per 1M at destination -9.242*** (0.768)

Origin is not developed X ERP use at destination 1.050** (0.504)

Constant -154.399*** (14.744)

-------------------------------------------------------------------------------

Observations 2,268

R2 0.419

Adjusted R2 0.407

Residual Std. Error 2.940 (df = 2220)

F Statistic 34.075*** (df = 47; 2220)

===============================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

The interaction effect between the origin being a developing country and the percentage of businesses using ERP (Enterprise Resource Planning) software at destination is a statically significant and positive determinant of international trade. Although the ERP development/support are not among the available services sectors, the ERP usage by businesses serves as a proxy of overall adoption of ICT technologies which make the international trade in services easier. In addition, such interaction effect describes both partners in some way. The ERP usage by business serves as an indirect indicator of the demand on ICT-related services at destination, while the `developing' status of origin could embed some traditional supply-side advantages of the developing countries (such as cost of labour). This goes in line with the hypothesis that the Internet affects developing and developed countries differently: the interaction effect correspond to the slope of the internet variable being different depending on the origin country being developing presumably due to developing countries export more services related to the Internet to the developed partners than other pairs of partners do.

To expose possible trends in time the model was re-estimated for the other years of the sample (2011-2015). This dataset of other years is used to further investigate the possible trade pattern `export of services from developing countries to the developed ones using the Internet and the interaction effects related to this pattern.

Data for the year 2011

Table 11. Model 9 estimation results.

==============================================================================

Dependent variable:

---------------------------

Trade flow

------------------------------------------------------------------------------

Predicted Probability of trade 0.155*** (0.022)

Distance -0.079 (0.064)

Common official language 1.283*** (0.365)

Contiguity -0.675** (0.295)

Mass variable 21.713*** (2.918)

Origin is developed 101.843*** (13.885)

Destination is developed 73.362*** (9.796)

Number of secure Internet servers per 1M at origin -54.270*** (7.331)

Percentage of Internet users at origin 86.397*** (10.665)

Constant -1,297.122*** (169.920)

------------------------------------------------------------------------------

Observations 2,655

R2 0.395

Adjusted R2 0.384

Residual Std. Error 3.091 (df = 2604)

F Statistic 34.024*** (df = 50; 2604)

==============================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

Model estimated on the data for the year 2011 is somewhat different from the model based on the year 2010 data. First, industries seem to become more similar - most of the industry dummies are insignificant. However, one more services sector has a significant (level 0.1) and positive coefficient - health services. In other means, the results are quite similar.

Table 12. Model 10 estimation results.

===============================================================================

Dependent variable:

-----------------------

Trade flow

-------------------------------------------------------------------------------

Predicted Probability of trade 0.155*** (0.022)

Distance -0.079 (0.064)

Common official language 1.283*** (0.365)

Contiguity -0.675** (0.295)

Mass variable 7.928*** (0.753)

Origin is developed 35.389*** (3.639)

Destination is developed 38.334*** (3.773)

Percentage of Internet users at origin 44.864*** (4.327)

Percentage of Internet users at destination 39.677*** (4.360)

Number of secure Internet servers per 1M at origin -18.671*** (1.839)

Number of secure Internet servers per 1M at destination -19.581*** (1.842)

Mobile cellular subscriptions per 100 at origin 6.488*** (1.169)

Mobile cellular subscriptions per 100 at destination 8.588*** (1.139)

Education industry X Internet users at origin 0.453* (0.246)

Construction industry X Internet users at origin -0.018 (0.246)

Telecom industry X Internet users at origin -0.144 (0.246)

Insurance industry X Internet users at origin -0.097 (0.246)

Financial industry X Internet users at origin -0.286 (0.246)

Health industry X Internet users at origin 0.488** (0.246)

Constant -677.802*** (66.180)

-------------------------------------------------------------------------------

Observations 2,655

R2 0.394

Adjusted R2 0.383

Residual Std. Error 3.093 (df = 2604)

F Statistic 33.927*** (df = 50; 2604)

===============================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

As with the data for year 2010, the result with interaction effects enabled is not qualitatively different from the result without it.

Data for the year 2012

For the year 2012 a special interaction effect was introduced to account for changes in data.

Table 13. Model 11 estimation results.

==========================================================================

Dependent variable:

---------------------------

Trade flow

--------------------------------------------------------------------------

Predicted Probability of trade 0.170*** (0.020)

Distance 0.021 (0.054)

Common official language 0.438 (0.279)

Contiguity 0.153 (0.277)

Mass variable 0.154** (0.064)

Destination is developed 1.866*** (0.424)

Origin is developed X Internet users at origin -0.758*** (0.079)

Constant -6.063* (3.566)

--------------------------------------------------------------------------

Observations 3,199

R2 0.382

Adjusted R2 0.372

Residual Std. Error 3.042 (df = 3144)

F Statistic 36.040*** (df = 54; 3144)

==========================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

The variable `Origin is developed X Internet users at origin' represents the interaction effect between origin being a developed country and the internet use at origin. As the coefficient for this interaction effect is statistically significant and negative, it could be inferred that interaction effect is larger when origin is a developing country (since every country is either developed or developing). This result suggests the existence of a special international trade pattern - export from the developing countries to the developed countries with the use of the Internet. The reason for this could probably be that the Internet lowers some information barriers to trade (i.e. makes the search for opportunities easier, simplifies the transactions and other operational processes related to services that previously required physical presence).

Table 14. Model 12 estimation results.

==========================================================================

Dependent variable:

---------------------------

Trade flow

--------------------------------------------------------------------------

Predicted Probability of trade 0.171*** (0.020)

Distance 0.022 (0.054)

Common official language 0.439 (0.279)

Contiguity 0.153 (0.277)

Mass variable 0.153** (0.064)

Destination is developed 1.867*** (0.424)

Origin is developed X Internet users at origin -0.758*** (0.079)

Education industry X Internet users at origin 0.374** (0.147)

Construction industry X Internet users at origin 0.051 (0.147)

Telecom industry X Internet users at origin -0.060 (0.148)

Insurance industry X Internet users at origin -0.190 (0.148)

Financial industry X Internet users at origin -0.318** (0.148)

Health industry X Internet users at origin 0.433*** (0.147)

Constant -5.993* (3.536)

--------------------------------------------------------------------------

Observations 3,199

R2 0.382

Adjusted R2 0.372

Residual Std. Error 3.042 (df = 3144)

F Statistic 36.049*** (df = 54; 3144)

==========================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

Again, analyzing the interaction effects does not yield qualitatively different results.

Data for the year 2013

Table 15. Model 13 estimation results.

==========================================================================

Dependent variable:

---------------------------

Trade flow

--------------------------------------------------------------------------

Predicted Probability of trade 0.033** (0.014)

Distance -0.033 (0.050)

Common official language 0.788*** (0.259)

Contiguity -0.262 (0.264)

Mass variable 0.180*** (0.061)

Destination is developed -0.445 (0.463)

Origin is developed X Internet users at origin -0.798*** (0.112)

Constant -5.733* (3.359)

--------------------------------------------------------------------------

Observations 3,434

R2 0.360

Adjusted R2 0.350

Residual Std. Error 2.987 (df = 3377)

F Statistic 33.956*** (df = 56; 3377)

==========================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

Year 2013 seem to display a change in trade patterns. Now almost every sectoral dummy variable is insignificant (except for the dummy of Human health services sector). In addition, the dummy variable `destination is developed' has also lost its significance casting doubt on the `export from developing to developed' trade pattern.

Table 16. Model 14 estimation results.

==========================================================================

Dependent variable:

---------------------------

Trade flow

--------------------------------------------------------------------------

Predicted Probability of trade 0.032** (0.014)

Distance -0.033 (0.050)

Common official language 0.787*** (0.259)

Contiguity -0.262 (0.264)

Mass variable 0.179*** (0.061)

Destination is developed -0.442 (0.463)

Origin is developed X Internet users at origin -0.798*** (0.112)

Education industry X Internet users at origin 0.202 (0.148)

Construction industry X Internet users at origin 0.125 (0.148)

Telecom industry X Internet users at origin 0.048 (0.148)

Insurance industry X Internet users at origin 0.019 (0.148)

Financial industry X Internet users at origin -0.217 (0.148)

Health industry X Internet users at origin 0.340** (0.148)

Constant -5.695* (3.337)

--------------------------------------------------------------------------

Observations 3,434

R2 0.360

Adjusted R2 0.350

Residual Std. Error 2.987 (df = 3377)

F Statistic 33.986*** (df = 56; 3377)

==========================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

Again, analyzing the interaction effects does not yield qualitatively different results.

Data for the year 2014

Table 17. Model 15 estimation results.

==========================================================================

Dependent variable:

---------------------------

Trade flow

--------------------------------------------------------------------------

Predicted Probability of trade 0.165*** (0.024)

Distance 0.0002 (0.064)

Common official language 1.757*** (0.372)

Contiguity -0.500 (0.307)

Mass variable 0.239*** (0.070)

Destination is developed -2.863*** (0.607)

Origin is developed X Internet users at origin -1.751*** (0.164)

Constant 2.719 (4.255)

--------------------------------------------------------------------------

Observations 2,870

R2 0.409

Adjusted R2 0.398

Residual Std. Error 3.195 (df = 2817)

F Statistic 37.421*** (df = 52; 2817)

==========================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

As for the year 2014, all the industry dummies are significant and negative (apparently except Information services sector which is the case where all dummies equals 0).

Table 18. Model 16 estimation results.

==========================================================================

Dependent variable:

---------------------------

Trade flow

--------------------------------------------------------------------------

Predicted Probability of trade 0.165*** (0.024)

Distance 0.00003 (0.064)

Common official language 1.757*** (0.372)

Contiguity -0.500 (0.307)

Mass variable 0.241*** (0.070)

Destination is developed -2.855*** (0.607)

Origin is developed X Internet users at origin -1.711*** (0.159)

Education industry X Internet users at origin -0.754*** (0.290)

Construction industry X Internet users at origin -0.913*** (0.294)

Telecom industry X Internet users at origin -0.830*** (0.295)

Insurance industry X Internet users at origin -1.057*** (0.297)

Financial industry X Internet users at origin -1.229*** (0.300)

Health industry X Internet users at origin -0.476* (0.288)

Constant 2.310 (4.197)

--------------------------------------------------------------------------

Observations 2,870

R2 0.408

Adjusted R2 0.397

Residual Std. Error 3.195 (df = 2817)

F Statistic 37.399*** (df = 52; 2817)

==========================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

The analysis of interaction effects yields qualitatively the same results.

Data for the year 2015

Table 19. Model 17 estimation results.

==========================================================================

Dependent variable:

---------------------------

Trade flow

--------------------------------------------------------------------------

Predicted Probability of trade 0.041** (0.019)

Distance -0.106 (0.065)

Common official language 1.145*** (0.370)

Contiguity -0.298 (0.339)

Mass variable -0.056 (0.069)

Destination is developed -1.285** (0.575)

Origin is developed X Internet users at origin -0.668*** (0.136)

Constant 9.709** (3.911)

--------------------------------------------------------------------------

Observations 2,370

R2 0.394

Adjusted R2 0.382

Residual Std. Error 3.104 (df = 2321)

F Statistic 31.486*** (df = 48; 2321)

==========================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

For the year 2015 the coefficients are somewhat unusual - many coefficients are insignificant (even the Mass variable) and further analysis is needed.

Table 20. Model 18 estimation results.

==========================================================================

Dependent variable:

---------------------------

Trade flow

--------------------------------------------------------------------------

Predicted Probability of trade 0.041** (0.019)

Distance -0.106 (0.065)

Common official language 1.145*** (0.370)

Contiguity -0.298 (0.339)

Mass variable -0.055 (0.069)

Destination is developed -1.277** (0.575)

Origin is developed X Internet users at origin -0.668*** (0.136)

Education industry X Internet users at origin -0.032 (0.244)

Construction industry X Internet users at origin -0.297 (0.244)

Telecom industry X Internet users at origin -0.150 (0.244)

Insurance industry X Internet users at origin -0.341 (0.244)

Financial industry X Internet users at origin -0.482** (0.245)

Health industry X Internet users at origin 0.159 (0.244)

Constant 9.617** (3.876)

--------------------------------------------------------------------------

Observations 2,370

R2 0.394

Adjusted R2 0.382

Residual Std. Error 3.104 (df = 2321)

F Statistic 31.461*** (df = 48; 2321)

==========================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

However, the analysis of interaction effect did not clarified the unusualness (although the dummy variable for financial services sector became significant).

Table 21. Model 19 estimation results.

==========================================================================

Dependent variable:

---------------------------

Trade flow

--------------------------------------------------------------------------

Predicted Probability of trade 0.055** (0.025)

Distance -0.109 (0.091)

Common official language 2.768*** (0.541)

Contiguity -1.120*** (0.358)

Mass variable 20.898*** (4.741)

Origin is developed X Internet users at origin -376.317*** (86.699)

% of Businesses with a website at destination 1,780.315*** (397.932)

% of Businesses with a website at origin 222.293*** (55.984)

% of Businesses using CRM at destination 560.147*** (121.541)

% of Businesses using ERP at destination 101.162*** (23.340)

Constant -10,471.140*** (2,350.667)

--------------------------------------------------------------------------

Observations 1,730

R2 0.434

Adjusted R2 0.420

Residual Std. Error 3.123 (df = 1687)

F Statistic 30.854*** (df = 42; 1687)

==========================================================================

Note: *p<0.1; **p<0.05; ***p<0.01

With introduction of some proxy-variables for using of technologies the estimation becomes more robust. The coefficients are significant and have the expected signs. The variables of use of CRM, ERP and websites by businesses serve as a proxies for technology usage (and history of such usage) which is necessary to trade services remotely via the Internet. However, the dummy variables for the services sectors are still insignificant.

Overall, the trend of vanishing the differences between sectors of services is present while the trade pattern of exporting from developing countries (to any countries) remains significant.

Conclusion & Policy implications

As the results of the research suggests, the origin's internet variables are significant determinant of trade flows while destination's internet variables are not. Therefore, the encouragement of the Internet's diffusion could be a possible strategy for the leaders of developing countries in particular for the goal of increasing nation's export and international trade.

I find that the Percentage of Internet users at origin is the accurate predictor of trade flows while the other more infrastructural indicators are not. That is why the Internet use should be used at a key metric for measuring policy's success.

The usage of some Internet-related technologies was found to influence the trade flows positively, so the countries with the higher rate of using the technologies (such as websites, ERP and CRM) are the main potential trade partners for the export of services.

I find that the services sectors most distinguished above average are traded via GATS mode 2 with the impact of the diffusion of the Internet. That means that the Internet seem to make some information barriers lower (i.e. search of information, negotiation, paying). On the other side, this type of services trade still involves the consumer to physically come to the origin country. Therefore, the transportation and overall safety for tourists should also be developed to attract the consumers of such services. However, a particular service sector should rather not be focused on, as the differentiation between the sectors seem to fade away over time.

Limitations

The main limitation of the research is the typical problem for such papers - the data insufficiency. This insufficiency spreads across all three dimensions of data. First, both backward and forward extension of the timeframe was found to be impossible due to issues with changing OECD membership, lack of recent data on `gravity' variables and lack of a more older data for `internet' variables. Second, the geographical extension of the data set is also hardly possible without having to combine several data sources due to an essential scarcity of the detailed data on trade flows by partner. Third, the sectoral extension of data set is associated with the need of translating one classification into an another one which is possible without introducing too strong assumptions only for a very limited number of sectors.

However, the best knowledge sources and recent papers were used to make proper corrections. The problem could only be fully addressed when (and if) more data will become publicly available.

Directions for future research

One of the directions of future research could be to validate the obtained results on a larger data set (in terms of time/countries/industries).

Another direction could be to construct a custom index of Internet's impact containing all the significant variables which could explain how the diffusion of the Internet influences international trade or explain how the Internet lowers the information barriers.

One more direction would be to research the demographic factors not considered herein: the human capital, i.e. the quality of STEM education and relative price of labour which could be the key factors for enabling export of Internet-related services from developing countries.

Another direction could be to determine why education and health services was so special service sector with the positive and significant effect but have lost its specialty over time.

Finally, the impact of the Internet on the firm level could be analyzed: the way the Internet makes export easier for a firm.

References

1. Anderson, J.E., 2011. The Gravity Model. The Annual Review of Economics, Vol. 3, 133-60.

2. Anderson, J.E., van Wincoop, E., 2003. Gravity with Gravitas: A Solution to the Border Puzzle. The American Economic Review, Vol. 93, No. 1, 170-192.

3. Anderson et al., 2018. Dark Costs, Missing Data: Shedding Some Light on Services Trade. European Economic Review, Vol. 105, 193-214.

4. Baier, S.L., Bergstrand, J.H., 2007. Do free trade agreements actually increase members' international trade? Journal of International Economics, Vol. 71, 72-95.

5. Baldwin R., Harrigan J., 2011. Zeroes, Quality, and Space: Trade Theory and Trade Evidence. American Economic Journal, Vol. 3, No. 2, 60-88.

6. Baldwin R., Taglioni D., 2006. Gravity for Dummies and Dummies for Gravity Equations. NBER Working Paper No. 12516.

7. Baltagi, B.H., Egger, P., Pfaffermayr, M., 2003. A generalized design for bilateral trade flow models. Economics Letters, Vol. 80, 391-397.

8. Bergstrand, J.H., 1985. The Gravity Equation in International Trade: Some Microeconomic Foundations and Empirical Evidence. The Review of Economics and Statistics, Vol. 67, No. 3, 474-481.

9. Bergstrand, J.H., 1989. The Generalized Gravity Equation, Monopolistic Competition, and the Factor-Proportions Theory in International Trade. The Review of Economics and Statistics, Vol. 71, No. 1, 143-153.

10. Bernard, A.B., Jensen, J.B., Redding, S.J., Schott, P.K., 2007. Firms in International Trade. Journal of Economic Perspectives, Vol. 21, 105-130.

11. Blum, B.S., Goldfarb, A., 2006. Does the internet defy the law of gravity? Journal of International Economics, Vol. 70, 384-405.

12. Brun, J., Carrere, C., Guillaumont, P., de Melo, J., 2005. Has Distance Died? Evidence from a Panel Gravity Model. The World Bank Economic Review, Vol. 19, No. 1, 99-120.

13. Chen N., Novy D., 2012. On the Measurement of Trade Costs: Direct vs. Indirect Approaches to Quantifying Standards and Technical Regulations. World Trade Review, Vol. 11, No. 3, 401-414.

14. Choi, C., 2010. The effect of the Internet on service trade. Economics Letters, Vol. 109, 102-104.

15. Clarke, G.R.G., Wallsten, S.J., 2006. Has the Internet Increased Trade? Developed and Developing country evidence. Economic Inquiry, Vol. 44, No. 3, 465-484.

16. Egger, P., Pfaffermayr, M., 2003. The proper panel econometric specification of the gravity equation: A three-way model with bilateral interaction effects. Empirical Economics, Vol. 28, 571-580.

17. Freund, C., Weinhold, D., 2002. The Internet and International Trade in Services. The Economics of Technology and Innovation, Vol. 92, No. 2, 236-240.

18. Helpman E., Melitz M., Rubinstein Y., 2008. Estimating Trade Flows: Trading Partners and Trading Volumes. The Quarterly Journal of Economics, Vol. 123, No. 2, 441-487.

19. Kepaptsoglou, K., Karlaftis, M.G., Tsamboulas, D., 2010. The gravity model specification for modeling international trade flows and free trade agreement effects: a 10-year review of empirical studies. The Open Economics Journal, Vol. 3, 1-13.

20. Kimura, F., Lee, H., 2006. The Gravity Equation in International Trade in Services. Review of World Economics, Vol. 142, No. 1, 92-121.

21. Matyas, L., 1997. Proper Econometric Specification of the Gravity Model. The World Economy, Vol. 20, No. 3, 363-368.

22. McCallum, J., 1995. National Borders Matter: Canada-U.S. Regional Trade Patterns. The American Economic Review, Vol. 85, No. 3, 615-623.

23. Meijers, H., 2014. Does the Internet generate economic growth, international trade, or both? International Economics and Economic Policy, Vol. 11, No. 1, 137-163.

24. Rueda-Cantuche J.M., Kerner R., Cernat L., Ritola V., 2016. Trade in Services by GATS Modes of Supply: Statistical Concepts and First EU Estimates. Chief Economist Note, Issue 3 (November 2016). Link: http://trade.ec.europa.eu/doclib/docs/2016/december/tradoc_155119.pdf

25. UN International Telecommunication Union, 2014. Measuring the Information Society Report 2014, pages 35-40. Link: https://www.itu.int/en/ITU-D/Statistics/Documents/publications/mis2014/MIS2014_without_Annex_4.pdf

26. Vemuri, V.K., Siddiqi, S., 2009. Impact of Commercialization of the Internet on International Trade: A Panel Study Using the Extended Gravity Model. The International Trade Journal, Vol. 23, No. 4, 458-484.

27. Wallsten, S., 2015. The Competitive Effects of the Sharing Economy: How is Uber Changing Taxis? Technology Policy Institute.

Appendix 1.

Results replication

Data processing and model estimation were both made with the use of R programming language. The raw datasets and source code are available by link: https://drive.google.com/drive/folders/1kYDmL-xqjMM7Cy7v3g3SxvOeCLDHYaCA?usp=sharing. There are two main ways of replicating the results. First way is to directly launch .R script (raw data needs to be placed at the same folder with the script). This way a whole replication of the results from scratch is possible at a cost of time and computing power. The second way of replicating the results is to use the final data frames (a special R data structure) without replicating the creation of these data frames. These data frames are available in .rda files corresponding to each year. It is possible to conduct a gravity model estimation on these data frames to replicate the results and/or to get other results with a different research goal in mind. This approach allows for a fast re-use of the data set. In addition, the underlying R functions in the script were designed to be generic and could be applied to another time-period/countries/sectors assuming the raw data formatting is unchanged. That is why the files associated with current text could leverage future research and provide reusability of data (source code was commented and the functions were annotated to make the code usage easier).

Размещено на Allbest.ru

...