Warsaw University Faculty of Economic Sciences Master of Arts in Development Economics

Elizabeth Rivard Album number: Joanna Nestorowicz album number: 257801

Michaù Grajek’s research on

Diffusion of ISO 9000 Standards and International Trade A critical review The author in his research explores the impact of ISO certification on trade and FDI flows between pair of countries. Unfortunately he lacks data on LDC-LDC ISO, investment and trade flows. This part of the analysis would have been a very important argument in the discussion on the relevance, importance and effectiveness of the ISO 9000 framework. Especially that ISO 9000 is being ‘advertised’ as a tool in creating an inclusive world economy. Having such developing and unstable, both politically and economically, states in mind, we propose to add an other independent variable into the regression – the Transparency International Corruption Perception Index (CPI). This factor might be very influential and might be a very strong signal for exporting and investing countries not to direct their actions at such countries. The negative signaling effect might be significantly weakening the positive signaling effect of ISO 9000 certification. Warsaw, March 2006

1

Contents Contents...................................................................................................................................... 2 Introduction ................................................................................................................................ 3 Literature review ........................................................................................................................ 4 Analysis of the theoretical framework ....................................................................................... 7 Analysis of the Statistical Methodology .................................................................................... 9 Analysis of the Empirical Findings .......................................................................................... 10 Conclusions .............................................................................................................................. 12 Extension proposal ................................................................................................................... 13 Theoretical Basis .................................................................................................................. 13 Description of Corruption Index .......................................................................................... 13 Statistical Methodology........................................................................................................ 14 Empirical Results ................................................................................................................. 14 General regression by FDI without Kraay Index ......................................................... 15 General regression by FDI with Kraay Index .............................................................. 16 Regression of DC-LDC pairs by FDI and exports without Kraay Index ..................... 17 Regression of DC-LDC pairs by FDI and exports with Kraay Index ......................... 18 Regression of DC-DC pairs by FDI and exports without Kraay Index ..................... 19 Regression of DC-DC pairs by FDI and exports with Kraay Index............................ 20 Conclusions .......................................................................................................................... 21 Bibliography............................................................................................................................. 22

2

Introduction The issue of standardization and certification is very widely discussed nowadays. Mainly because of the growing number of tolls and mechanism offered in such work frames. These mechanisms have both strong supporters and a strong opposition. Examining the effectiveness and externalities arising form existing mechanisms would allow policy makers to decide in which way they should proceed – create, enhance, promote and adopt such standardization mechanisms, or - on the contrary – discourage domestic and foreign companies to implement such measures when operating on the internal and international markets.

In the first part of this paper we will discuss the theory behind the presented research – it’s validity and applicability to the topic. Later we will examine the methodology used to conduct this research. To summarize the first part of the critique we will also discuss the findings of the presented paper.

The last part will be more constructive and will include a proposal of an extension to the research under discussion. We will try to examine whether corruption, measured by the Kraay Index is a significant factor in terms of exports or FDI flows to a certain country. We will assign the Kraay index only to the recipient country as this relation is the one we are especially interested in. Comparing the result of a regression with and without the Kraay Index included should tell us whether it is a factor worth further studies or not. Our intuition is that the level of corruption should be negatively correlated with export and FDI flows.

3

Literature review The topic of ISO 9000 is widely discussed although literature dealing with the impact of standardization on trade is rather scarce. Most papers focus on the motivations of company managers to invest in ISO certification, they also deal with marginal returns following ISO 9000 adoption.

In one research Anderson, Daly and Johnson (1999) explore whether company managers would apply for ISO 9000 certification mainly due to the need to comply with regulator or customer requirements or whether ISO 9000 is generally a component of total quality management policies (TQM). The null- hypothesis that firms obtain ISO 9000 certification primarily to comply with government and customer demands.

It ay be hard to conclude which of these two factors driving companies to obtain ISO 9000 would have what kind of impact on international trade. It may be that if entering a foreign market requires ISO 9000 certification (as it happens in the EU in many cases) then the possibility of obtaining such a certificate would enhance trade, or - as a matter of fact make it possible at all. But it may be also that a foreign government or client does not require such certification, but they do care about quality in general. In this case ISO 9000 might be treated not as a formal requirement but as a tool for the companies managers to obtain higher quality of their products, what could result in greater competitiveness in a foreign market and therefore encourage the company to go for it.

One of the findings of the study is that:

“Sales in international markets positively influence the decision to seek ISO 9000. Consistent with the greater role that ISO 9000 has in the EC, sales in Europe (EUR) play a greater role in the ISO 9000 adoption decision than do sales in other international markets ( INTNL) “. The study also signals that ISO-demanding regulation affects potential market entrants more, than companies which had previous market presence. An other important remark is that ISO 9000 standards do not necessarily indicate the highest possible level of effectiveness and quality: 4

“ISO 9000 may not guarantee that the firm has the most effective quality management program. However, the structure of the ISO 9000 standards suggest that certification indicates attainment of an effective threshold of quality assurance.” What they have also indicated is that in some cases, mainly of firms which have a nongovernmental (the study dealt with U.S. companies’ certification) major customer, there are other forms of signaling good quality then ISO certification. Yet on the other hand it is said that they found no evidence that companies for which the government is a major client would tend to seek ISO certification as well. For whom is it then?

“Firms with a small number of large customers are less likely to obtain certification because better alternatives for communicating and contracting for quality exist. […] Firms with alternative mechanisms for signaling quality (e.g., branded consumer goods) or for which the particular form of quality control embedded in the ISO 9000 standards may be inappropriate (e.g., industries with a high level of process or product innovation) are unlikely to obtain certification.” Finally they conclude:

“Contrary to the view held by many critics of the ISO 9000 quality assurance standards, certification is not sought primarily in response to regulatory requirements. We find strong evidence to support proponents’ claims that managers are obtaining ISO 9000 certification as a credible public signal of effective quality management practices. This signal is sought even when other public signals of quality attainment are present.

[…]

In summary, the results indicate that customer and regulatory compliance are inadequate explanations for the widespread adoption of ISO 9000 in North 5

American manufacturing firms. Managers obtain certification as a means of providing credible signals of quality assurance to external parties, and ISO 9000 complements rather than substitutes for more developed total quality management efforts. Thus, ISO 9000 is being adopted as one tool in a larger strategy of achieving

competitive

advantage

through

quality

management

and

communicating quality results. For most firms, complying with customer or regulatory requirements appears to be a secondary consideration.” The fact that there may be other, lower cost mechanisms which reduce transactions costs. Moreover it is argued that codification of existing practices may discourage from continuous improvement efforts. This issue might be of high importance and relevance especially for the developing economies.

6

Analysis of the theoretical framework As it is said in other research (Chemical Week 1993) obtaining an ISO 9000 certificate at a manufacturing site usually takes from nine to 28 months. If such a manufacturing facility employs 100 people and has few procedures which assure quality, it may expect to spend about $50,000 for system development and registration (Arnold 1994). For larger firms (sales from $100-500 million) the average total one-time cost of registration (excluding further review audits) was $300,000. The same firms reported average annual savings of $200,000. This means that an average payback period for ISO certification is 1.5 years. However, this did show dramatic differences across the sample; the median payback was 3.3 years, the first quartile of the companies occurred at 1.2 years and the third quartile at 21.4 years. This would mean that analyzing single companies might be of better use in terms of information

An other issue which is hard measure, but which clearly affects the ISO 9000 does not proxy for quality of design, the aspect of product quality typically associated with aesthetics or functionality. Such a factor as aesthetics would affect the demand for a certain product, yet it does not ‘show’ when only ISO certification is taken into consideration. It might have been useful to compare the different situations – progressive (as assumed) economic development within the time span of developing a ISO standardization structure and further attainment of certificates by more and more companies.

In his theoretical analysis the author gave a lot of significance to the quantity of ISO certificates in a certain country. The issue though is that with the presence of transnational corporations a product, made up of several components, each of which being produced in a different country, has to have ISO certification for each of these parts. Such procedures enforce the adoption of ISO certification on the manufacturer. This is a much different situation then one in which the ISO infrastructure is being adapted for the purely domestic companies. This effect could be captured if the ISO data would be disaggregated into specific brands, industries or companies. Unfortunately, predictably, such data is not available, yet this would add one more motivational effect into the theory presented by the author. The theory makes an optimistic assumption as well, that ISO certification is something consciously ‘wanted’ by a certain company. But in the light of the approach presented above it might be

7

merely ‘less evil’. Just as the situation is with joining any other inter- or supra national structures, frameworks and organizations.

Some motivational strategies of implementing ISO mentioned by the authors of the reviewed literature are the following:

8

Analysis of the Statistical Methodology

“Employed panel data on trade and FDI reported by OECD nations and then test the impact of ISO 9000 adoption on country-pair economic relationships.” (3) The authors acknowledge the issue of complementarity between trade and FDI and attempt to reconcile the econometric difficulties associated with this type of relationship. Although FDI and trade are not dependent on one another, they are determined, in most cases, by similar factors. Thus, both the FDI and export equations will have the same regressors and the authors employ a SUR (Seemingly Unrelated Regression) model in order to compensate for the relationship between these two variables. (word?)

Essentially, the methodology employed is sound and consistent with the data available and generally accepted econometric principles. When considering the explanatory variables controlling for standard export and FDI determinants, the infrastructure variable, as captured through airline passengers carried, seems to be much less robust in its explanatory power. Although this may be a somewhat useful variable when considering trade flows between developed countries, it is much less useful in the case of flows from developed to developing countries and vice versa.

One could argue that it is possible that the infrastructure of

developing countries may be significantly underestimated or overestimated, which could be particularly relevant when considering countries that have large cities located on the coast, allowing access to trade via the sea, thus underestimating the infrastructure.

Likewise,

countries with developed tourism infrastructure may skew the results by overestimating this variable since many countries, particularly in the developing world, have high numbers of tourists but the majority of the country may not have developed infrastructure or be attractive to non-tourism investment or FDI.

(The infrastructure variable was inconsistent for all

country-pair types except LDC-DC. p.29)

However, despite the measures taken to limit the endogeneity between FDI and trade flow variables and in particular the error terms, it is likely that factors affecting these areas are unaccounted for. Finally, the authors did not consider the timing of ISO 9000 adoption decisions and FDI/trade flows, such that it is unable to be determined whether ISO came before or after an increase in FDI/trade. Further, since ISO adoption requires a certain level of revenue to be available, it may be the case the firms who have revenue to update their 9

processes are the same firms exporting a large portion of their product output, thus it may not necessarily contribute to increased trade.

Finally, as the authors note in their empirical findings section, there appeared to be a collinearity problem between non-tariff barriers and capital-market access. This seems like an obvious discrepancy given that countries with low non-tariff barriers are likely to allow relatively open access to their capital markets and investment in general.

Analysis of the Empirical Findings

“Proposition 1: The diffusion of ISO 9000 certification in both home-nations and foreign-nations generates enhanced exports within country-pair economic relationships; however, for the home-effect (selling-end) will be more robust than the host-effect (buying-end).”

“Proposition 2: The diffusion of ISO 9000 certification in both home-nation and foreign-nations generates enhanced FDI within country-pair economic relationships; however, the host-effect (selling-end) will be more robust than the home-effect (buyingend).”

The authors concluded that the validity of the above propositions is supported by their analysis.

“Proposition 3: The impact of ISO 9000 diffusion on both trade and FDI is likely to be greatest in developing nations as compared to developed nations.”

Interestingly, ISO 9000 certification appears to have no effect on developed-developed country trade and investment relationships.

The authors had several interesting findings, most of which were consistent with the predicted outcomes of statistical analysis.

10

The fact that ISO certification appeared to have no significant affect in explaining developed-developed country trading leads one to question the extent that it affects developing-developed trading, which the authors conclude is affected by ISO adoption in the developing nations. It seems likely the difference between developed-developing country trade/FDI flows could be a function of one or more variables other than ISO 9000, particularly given that the firms that are likely to adopt ISO standards would likely have other common characteristics that affect the attractiveness of their exports and/or their investment potential of the country in which they are located.

Also, the authors admit that the export equation variables were unreliable due to severe endogeneity issues that were unable to be rectified. Although they claim that the ISO variable equations were not similarly affected due to their exogenous nature, more robust testing of these variables is in order to ensure that important variables have not been excluded.

11

Conclusions Although the authors appear to offer reliable insight into the nature of ISO 9000 diffusion and its effect on FDI and trade between developing and developed countries, in reality, it appears that their hypotheses are extremely difficult to verify given the constraints on the data and the ability to control for endogenous variables. Most importantly, the authors did not successfully control for the relationship between FDI and exports to ISO 9000 certification, such that it is likely that FDI flows could be a determinant of developing countries to adopt ISO certification.

There are still numerous factors, such the obligatory implementation of ISO standards in some LDCs due to the production of a mere part of a product which has to be ISO certified as a whole.

12

Extension proposal

Theoretical Basis Although the author of this paper took into account a large number of variables that may influence FDI and export flows. However, one important variable that may be important in this context is the level of perceived and/or real corruption, particularly in the recipient country. We would assume that corruption in the country from which the FDI and exports are flowing would not be important, particularly since only developed countries (DC) were tested by the author and that one could posit that the level of corruption in these countries is minimal, at least compared to less-developed countries. This hypothesis was confirmed by the data.

Although there is still some debate on how much the perceived level of corruption in a country influences its business relationships with outsiders, particularly in the context of FDI, a substantial body of literature concluded that corruption has a significant influence on the flow of FDI to a host country.1

Description of Corruption Index

“Control of corruption (CC), the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as “capture” of the state by elites and private interests.” (Kraay 4) Scores lie between a high of 2.5 and a low of -2.5, and we have used the actual values reported, rather than a ranking or dummy variable, for example. Barbados and Brazil were the only countries with no observations and a few others2 had only a partial list of observations for our time period.

Although this index only began to be compiled annually in 2001, we took an average of the years to compute the index for the missing years of data. This approximation of the data should not have a significant impact on our results due to the fact that the value of the 1

Habib and Zurawicki 2002, Wei and Shleifer 2000 Countries with missing years of data were: Belize, Dominica, Grenada, Macao, St. Lucia, Seychelles, and Swaziland. 2

13

index did not change drastically for any country during the time period considered here. Exceptions included Indonesia, Israel, and Paraguay which experienced significant increases in corruption, while Estonia and Qatar had a rather marked decrease.

Statistical Methodology Although it was not possible to re-create the regressions done by the author exactly, we attempted to simulate his work as much as possible. We also ran regressions using only the data the author’s specified and then again with the Kraay Index added in order to have a direct comparison of the results.

H0: The level of corruption in the recipient country does not impact the level of exports and/or FDI. H1: The level of corruption in the recipient country does impact the level of exports and/or FDI.

The hypothesis was tested on two different types of country pairs, DC-LDC and DCDC. Due to our restricted set of data, we were not able to test LDC-DC flows as the author did in his research.

Empirical Results According to the tests we ran, the level of corruption, as measured by the Kraay Index, was not significant in influencing exports or FDI in either of the country pair types.

14

General regression by FDI without Kraay Index regress fdioutpos rprt_gdp prtn_gdp rprt_iso prtn_iso prtn_pass prtn_reer rprt_reer prtn_smi rprt_smi prtn_polcon prtn_tci prtn_tariffs prtn_fcm prtn_hf Source | SS df MS -------------+-----------------------------Model | 1.0160e+11 14 7.2568e+09 Residual | 2.2904e+11 1324 172987336 -------------+-----------------------------Total | 3.3063e+11 1338 247108365

Number of obs = 1339 F( 14, 1324) = 41.95 Prob > F = 0.0000 R-squared = 0.3073 Adj R-squared = 0.3000 Root MSE = 13152

-----------------------------------------------------------------------------fdioutpos | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------+---------------------------------------------------------------rprt_gdp | .0008546 .0005302 1.61 0.107 -.0001855 .0018946 prtn_gdp | -.0026352 .0020933 -1.26 0.208 -.0067418 .0014714 rprt_iso | .03672 .0232124 1.58 0.114 -.0088171 .0822572 prtn_iso | .3042365 .0721144 4.22 0.000 .1627655 .4457075 prtn_pass | .0316213 .0647532 0.49 0.625 -.0954087 .1586513 prtn_reer | 26.39047 54.17679 0.49 0.626 -79.89125 132.6722 rprt_reer | 31.83658 55.90314 0.57 0.569 -77.83182 141.505 prtn_smi | .0031892 .0019025 1.68 0.094 -.0005431 .0069214 rprt_smi | .001245 .0003731 3.34 0.001 .0005131 .0019769 prtn_polcon | 2017.109 3535.755 0.57 0.568 -4919.184 8953.403 prtn_tci | 296.6643 731.0964 0.41 0.685 -1137.569 1730.898 prtn_tariffs | -16979.02 23459.16 -0.72 0.469 -63000.19 29042.15 prtn_fcm | -115.2113 431.8064 -0.27 0.790 -962.3107 731.8881 prtn_hf | 362.6348 387.3459 0.94 0.349 -397.244 1122.513 _cons | -11807.86 8478.89 -1.39 0.164 -28441.38 4825.669

15

General regression by FDI with Kraay Index

regress fdioutpos prtn_Kraay rprt_gdp prtn_gdp rprt_iso prtn_iso prtn_pass prtn_reer rprt_reer prtn_smi rprt_smi prtn_polcon prtn_tci prtn_tariffs prtn_fcm prtn_hf

Source | SS df MS -------------+-----------------------------Model | 1.0160e+11 15 6.7731e+09 Residual | 2.2903e+11 1323 173117883 -------------+-----------------------------Total | 3.3063e+11 1338 247108365

Number of obs = 1339 F( 15, 1323) = 39.12 Prob > F = 0.0000 R-squared = 0.3073 Adj R-squared = 0.2994 Root MSE = 13157

-----------------------------------------------------------------------------fdioutpos | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------+---------------------------------------------------------------prtn_Kraay | -2.879043 72.35374 -0.04 0.968 -144.8196 139.0615 rprt_gdp | .000854 .0005306 1.61 0.108 -.0001868 .0018948 prtn_gdp | -.0026363 .0020943 -1.26 0.208 -.0067448 .0014722 rprt_iso | .0367288 .0232222 1.58 0.114 -.0088276 .0822852 prtn_iso | .304326 .0721766 4.22 0.000 .1627328 .4459191 prtn_pass | .0315533 .0648001 0.49 0.626 -.0955688 .1586755 prtn_reer | 26.36115 54.20223 0.49 0.627 -79.97055 132.6929 rprt_reer | 31.76596 55.95238 0.57 0.570 -77.99911 141.531 prtn_smi | .0031902 .0019034 1.68 0.094 -.0005438 .0069242 rprt_smi | .0012453 .0003733 3.34 0.001 .000513 .0019776 prtn_polcon | 2031.185 3554.733 0.57 0.568 -4942.344 9004.714 prtn_tci | 293.3426 736.1209 0.40 0.690 -1150.749 1737.434 prtn_tariffs | -17142.29 23824.02 -0.72 0.472 -63879.27 29594.69 prtn_fcm | -115.5216 432.0397 -0.27 0.789 -963.0793 732.036 prtn_hf | 362.0417 387.7786 0.93 0.351 -398.6864 1122.77 _cons | -11778.93 8513.193 -1.38 0.167 -28479.76 4921.905

16

Regression of DC-LDC pairs by FDI and exports without Kraay Index -> DC_LDCdummy = 0 Source | SS df MS Number of obs = 615 ------------+-----------------------------F( 14, 600) = 26.56 Model | 1.7014e+09 14 121525531 Prob > F = 0.0000 Residual | 2.7449e+09 600 4574813.85 R-squared = 0.3827 ------------+-----------------------------Adj R-squared = 0.3682 Total | 4.4462e+09 614 7241442.58 Root MSE = 2138.9 -----------------------------------------------------------------------------fdioutpos | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------+---------------------------------------------------------------rprt_gdp | .0005435 .0001335 4.07 0.000 .0002812 .0008058 prtn_gdp | .003699 .0016651 2.22 0.027 .0004288 .0069692 rprt_iso | .0171731 .005418 3.17 0.002 .0065327 .0278136 prtn_iso | -.1222917 .0697636 -1.75 0.080 -.2593022 .0147188 rprt_reer | -15.94356 14.11263 -1.13 0.259 -43.65971 11.77258 prtn_reer | 1.453528 12.2814 0.12 0.906 -22.66622 25.57327 prtn_pass | -.0100653 .0321242 -0.31 0.754 -.0731549 .0530243 prtn_polcon | 49.58321 1363.431 0.04 0.971 -2628.094 2727.261 prtn_smi | .0027807 .0018509 1.50 0.134 -.0008544 .0064157 rprt_smi | 9.95e-07 .0000935 0.01 0.992 -.0001826 .0001846 prtn_tariffs | -908.037 4477.725 -0.20 0.839 -9701.955 7885.881 prtn_tci | 718.5609 174.2633 4.12 0.000 376.3208 1060.801 prtn_fcm | 172.9511 91.71629 1.89 0.060 -7.172898 353.075 prtn_hf | 216.4572 109.6784 1.97 0.049 1.057032 431.8573 _cons | -3592.372 2079.704 -1.73 0.085 -7676.756 492.0117 -> DC_LDCdummy = 0 Source | SS df MS Number of obs = 72 ------------+-----------------------------F( 18, 53) = 21.72 Model | 836904099 18 46494672.2 Prob > F = 0.0000 Residual | 113458487 53 2140726.17 R-squared = 0.8806 ------------+-----------------------------Adj R-squared = 0.8401 Total | 950362587 71 13385388.5 Root MSE = 1463.1 -----------------------------------------------------------------------------exports | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------+---------------------------------------------------------------rprt_gdp | .0027691 .0005085 5.45 0.000 .0017492 .003789 prtn_gdp | 1.10583 3.163209 0.35 0.728 -5.238765 7.450424 rprt_iso | .0335589 .0137434 2.44 0.018 .0059931 .0611247 prtn_iso | 12.35218 35.2367 0.35 0.727 -58.32368 83.02805 fdioutpos | 1.026244 .2032038 5.05 0.000 .6186689 1.433819 fdiroutpos | -.1612768 .1277903 -1.26 0.212 -.4175916 .0950381 fdiinpos | 11.70115 5.498469 2.13 0.038 .6726144 22.72968 fdirinpos | 10.76257 8.399111 1.28 0.206 -6.083912 27.60906 prtn_pass | -72.56263 208.9423 -0.35 0.730 -491.6479 346.5226 rprt_reer | 68.75078 60.24824 1.14 0.259 -52.09188 189.5934 prtn_reer | 105.7316 374.3293 0.28 0.779 -645.0779 856.541 prtn_smi | -2.922241 8.306895 -0.35 0.726 -19.58376 13.73928 rprt_smi | -.0022869 .0004032 -5.67 0.000 -.0030957 -.0014781 prtn_tci | 58947.02 169130.2 0.35 0.729 -280285.2 398179.2 prtn_polcon | -1031096 2956421 -0.35 0.729 -6960925 4898734 prtn_tariffs | -582994.1 1597133 -0.37 0.717 -3786438 2620450 prtn_fcm | -26063.33 74265.78 -0.35 0.727 -175021.6 122895 prtn_hf | 39481.83 112521.8 0.35 0.727 -186208.4 265172 _cons | 474837.1 1377807 0.34 0.732 -2288693 3238367

17

Regression of DC-LDC pairs by FDI and exports with Kraay Index -> DC_LDCdummy = 0 Source | SS df MS Number of obs = 615 ------------+-----------------------------F( 15, 599) = 24.76 Model | 1.7015e+09 15 113434069 Prob > F = 0.0000 Residual | 2.7447e+09 599 4582194.85 R-squared = 0.3827 ------------+-----------------------------Adj R-squared = 0.3672 Total | 4.4462e+09 614 7241442.58 Root MSE = 2140.6 -----------------------------------------------------------------------------fdioutpos | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------+---------------------------------------------------------------DC_LDCdummy | (dropped) prtn_Kraay | 2.210159 12.07195 0.18 0.855 -21.49834 25.91866 rprt_gdp | .0005444 .0001337 4.07 0.000 .0002817 .000807 prtn_gdp | .0037114 .0016679 2.23 0.026 .0004359 .006987 rprt_iso | .0171701 .0054224 3.17 0.002 .0065209 .0278192 prtn_iso | -.1221858 .0698222 -1.75 0.081 -.2593119 .0149404 prtn_pass | -.0104211 .0322088 -0.32 0.746 -.0736771 .0528348 prtn_reer | 1.401437 12.29459 0.11 0.909 -22.74431 25.54718 rprt_reer | -15.84184 14.13493 -1.12 0.263 -43.60188 11.91821 prtn_smi | .002813 .0018608 1.51 0.131 -.0008415 .0064675 rprt_smi | 5.05e-07 .0000936 0.01 0.996 -.0001834 .0001844 prtn_polcon | 29.00215 1369.153 0.02 0.983 -2659.922 2717.926 prtn_tci | 722.7489 175.8975 4.11 0.000 377.2981 1068.2 prtn_tariffs | -771.6545 4542.827 -0.17 0.865 -9693.46 8150.151 prtn_fcm | 174.3607 92.1126 1.89 0.059 -6.5422 355.2636 prtn_hf | 216.2759 109.7713 1.97 0.049 .6925825 431.8593 _cons | -3617.189 2085.79 -1.73 0.083 -7713.54 479.1614 -> DC_LDCdummy = 0 Source | SS df MS Number of obs = 72 ------------+-----------------------------F( 18, 53) = 21.72 Model | 836904099 18 46494672.2 Prob > F = 0.0000 Residual | 113458487 53 2140726.17 R-squared = 0.8806 ------------+-----------------------------Adj R-squared = 0.8401 Total | 950362587 71 13385388.5 Root MSE = 1463.1 -----------------------------------------------------------------------------exports | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------+---------------------------------------------------------------prtn_Kraay | -57.558 165.0338 -0.35 0.729 -388.574 273.458 rprt_gdp | .0027691 .0005085 5.45 0.000 .0017492 .003789 prtn_gdp | -.0153354 .0616951 -0.25 0.805 -.1390802 .1084094 rprt_iso | .0335589 .0137434 2.44 0.018 .0059931 .0611247 prtn_iso | -.2238275 .9494819 -0.24 0.815 -2.128247 1.680592 fdioutpos | 1.026244 .2032038 5.05 0.000 .6186689 1.433819 fdiroutpos | -.1612768 .1277903 -1.26 0.212 -.4175916 .0950381 fdiinpos | 11.70115 5.498469 2.13 0.038 .6726144 22.72968 fdirinpos | 10.76257 8.399111 1.28 0.206 -6.083912 27.60906 prtn_pass | -.2995123 2.445869 -0.12 0.903 -5.205305 4.60628 rprt_reer | 68.75078 60.24824 1.14 0.259 -52.09188 189.5934 prtn_reer | 208.1987 661.1742 0.31 0.754 -1117.949 1534.346 prtn_smi | .0463062 .2329165 0.20 0.843 -.4208653 .5134776 rprt_smi | -.0022869 .0004032 -5.67 0.000 -.0030957 -.0014781 prtn_tci | -1679.797 4991.217 -0.34 0.738 -11690.91 8331.317 prtn_tariffs | -8137.025 61356.02 -0.13 0.895 -131201.6 114927.6 prtn_fcm | -92.64532 837.1106 -0.11 0.912 -1771.676 1586.386 prtn_hf | 1606.447 4091.528 0.39 0.696 -6600.118 9813.013 _cons | -23869.12 54187.48 -0.44 0.661 -132555.4 84817.2

18

Regression of DC-DC pairs by FDI and exports without Kraay Index -> DC_LDCdummy = 1 Source | SS df MS Number of obs = 724 ------------+-----------------------------F( 14, 709) = 32.23 Model | 1.2235e+11 14 8.7391e+09 Prob > F = 0.0000 Residual | 1.9226e+11 709 271164147 R-squared = 0.3889 ------------+-----------------------------Adj R-squared = 0.3768 Total | 3.1460e+11 723 435135578 Root MSE = 16467 -----------------------------------------------------------------------------fdioutpos | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------+---------------------------------------------------------------rprt_gdp | .0002335 .0009065 0.26 0.797 -.0015461 .0020132 prtn_gdp | -.0011018 .0038128 -0.29 0.773 -.0085875 .0063839 rprt_iso | .076979 .0409631 1.88 0.061 -.0034445 .1574024 prtn_iso | .4387508 .1332708 3.29 0.001 .1770982 .7004034 rprt_reer | 34.97869 92.75478 0.38 0.706 -147.1282 217.0856 prtn_reer | 209.0869 131.8828 1.59 0.113 -49.84065 468.0145 prtn_pass | -.0155401 .1254573 -0.12 0.901 -.2618525 .2307722 prtn_polcon | 30914.86 20619.48 1.50 0.134 -9567.693 71397.41 prtn_smi | .001291 .0032392 0.40 0.690 -.0050685 .0076504 rprt_smi | .0028742 .0006383 4.50 0.000 .0016211 .0041274 prtn_tariffs | -274725.4 107496.3 -2.56 0.011 -485774.6 -63676.25 prtn_tci | -470.0414 2309.641 -0.20 0.839 -5004.596 4064.513 prtn_fcm | -5364.657 1801.38 -2.98 0.003 -8901.334 -1827.979 prtn_hf | 1135.278 906.3517 1.25 0.211 -644.1765 2914.732 _cons | -17914.8 30378.11 -0.59 0.556 -77556.62 41727.01 -> DC_LDCdummy = 1 Source | SS df MS Number of obs = 385 ------------+-----------------------------F( 18, 366) = 26.02 Model | 8.3010e+10 18 4.6117e+09 Prob > F = 0.0000 Residual | 6.4878e+10 366 177263586 R-squared = 0.5613 ------------+-----------------------------Adj R-squared = 0.5397 Total | 1.4789e+11 384 385126307 Root MSE = 13314 -----------------------------------------------------------------------------exports | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------+---------------------------------------------------------------rprt_gdp | .0029339 .001054 2.78 0.006 .0008612 .0050066 prtn_gdp | .013532 .0061874 2.19 0.029 .0013646 .0256993 rprt_iso | .0080524 .0425879 0.19 0.850 -.0756953 .0918 prtn_iso | .2687984 .195477 1.38 0.170 -.1156006 .6531974 fdioutpos | -.2398647 .1187469 -2.02 0.044 -.4733766 -.0063528 fdiroutpos | .8203955 .1125425 7.29 0.000 .5990845 1.041707 fdiinpos | .187405 .0704583 2.66 0.008 .0488512 .3259588 fdirinpos | -.1508399 .0415762 -3.63 0.000 -.232598 -.0690817 prtn_pass | -.5764736 .2751387 -2.10 0.037 -1.117525 -.0354226 rprt_reer | 78.71187 117.8053 0.67 0.504 -152.9482 310.372 prtn_reer | -126.8657 148.1039 -0.86 0.392 -418.107 164.3757 prtn_smi | -.0003875 .0034857 -0.11 0.912 -.007242 .0064669 rprt_smi | -.0012746 .0007424 -1.72 0.087 -.0027346 .0001853 prtn_tci | -1607.034 2810.297 -0.57 0.568 -7133.388 3919.321 prtn_polcon | 36620.37 22585.19 1.62 0.106 -7792.656 81033.4 prtn_tariffs | -176628.5 151467.5 -1.17 0.244 -474484.3 121227.2 prtn_fcm | -1237.758 2714.625 -0.46 0.649 -6575.977 4100.462 prtn_hf | 916.0737 1166.242 0.79 0.433 -1377.303 3209.451 _cons | -6198.27 32882.09 -0.19 0.851 -70859.82 58463.27

19

Regression of DC-DC pairs by FDI and exports with Kraay Index -> DC_LDCdummy = 1 Source | SS df MS Number of obs = 724 ------------+-----------------------------F( 15, 708) = 30.08 Model | 1.2246e+11 15 8.1643e+09 Prob > F = 0.0000 Residual | 1.9214e+11 708 271382828 R-squared = 0.3893 ------------+-----------------------------Adj R-squared = 0.3763 Total | 3.1460e+11 723 435135578 Root MSE = 16474 -----------------------------------------------------------------------------fdioutpos | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------+---------------------------------------------------------------prtn_Kraay | -597.0225 911.8459 -0.65 0.513 -2387.268 1193.223 rprt_gdp | .0002028 .000908 0.22 0.823 -.0015799 .0019856 prtn_gdp | -.001076 .0038145 -0.28 0.778 -.0085651 .0064131 rprt_iso | .0785692 .0410515 1.91 0.056 -.002028 .1591665 prtn_iso | .4397309 .1333329 3.30 0.001 .1779557 .701506 rprt_reer | 31.17159 92.97418 0.34 0.738 -151.3665 213.7097 prtn_reer | 208.3629 131.9406 1.58 0.115 -50.67875 467.4046 prtn_pass | -.0244917 .1262504 -0.19 0.846 -.2723616 .2233782 prtn_polcon | 31891.86 20681.7 1.54 0.124 -8712.934 72496.66 prtn_smi | .0014625 .003251 0.45 0.653 -.0049203 .0078453 rprt_smi | .0028948 .0006393 4.53 0.000 .0016397 .00415 prtn_tariffs | -282516.5 108196 -2.61 0.009 -494939.9 -70093.09 prtn_tci | -627.2685 2323.018 -0.27 0.787 -5188.096 3933.559 prtn_fcm | -4560.503 2180.841 -2.09 0.037 -8842.193 -278.8127 prtn_hf | 1098.697 908.4369 1.21 0.227 -684.8557 2882.249 _cons | -21175.82 30795.79 -0.69 0.492 -81637.81 39286.17 -> DC_LDCdummy = 1 Source | SS df MS Number of obs = 385 ------------+-----------------------------F( 19, 365) = 24.59 Model | 8.3030e+10 19 4.3700e+09 Prob > F = 0.0000 Residual | 6.4859e+10 365 177694856 R-squared = 0.5614 ------------+-----------------------------Adj R-squared = 0.5386 Total | 1.4789e+11 384 385126307 Root MSE = 13330 -----------------------------------------------------------------------------exports | Coef. Std. Err. t P>|t| [95% Conf. Interval] ------------+---------------------------------------------------------------prtn_Kraay | -345.5355 1033.832 -0.33 0.738 -2378.549 1687.479 rprt_gdp | .0029218 .0010559 2.77 0.006 .0008453 .0049983 prtn_gdp | .0132015 .0062733 2.10 0.036 .0008651 .025538 rprt_iso | .0093405 .0428134 0.22 0.827 -.0748515 .0935324 prtn_iso | .2579232 .198401 1.30 0.194 -.1322294 .6480758 fdioutpos | -.2411674 .1189552 -2.03 0.043 -.4750909 -.0072439 fdiroutpos | .8211453 .1127016 7.29 0.000 .5995193 1.042771 fdiinpos | .1872838 .0705448 2.65 0.008 .0485585 .3260091 fdirinpos | -.150693 .041629 -3.62 0.000 -.2325559 -.0688301 prtn_pass | -.561662 .279015 -2.01 0.045 -1.110341 -.0129833 rprt_reer | 78.47117 117.9507 0.67 0.506 -153.477 310.4193 prtn_reer | -122.8578 148.768 -0.83 0.409 -415.4078 169.6921 prtn_smi | -.0003158 .0034965 -0.09 0.928 -.0071916 .00656 rprt_smi | -.0012621 .0007443 -1.70 0.091 -.0027257 .0002015 prtn_tci | -1606.334 2813.714 -0.57 0.568 -7139.459 3926.791 prtn_polcon | 37634.1 22815.15 1.65 0.100 -7231.548 82499.75 prtn_tariffs | -190633.1 157333.9 -1.21 0.226 -500027.7 118761.4 prtn_fcm | -793.6636 3025.327 -0.26 0.793 -6742.922 5155.594 prtn_hf | 945.7693 1171.036 0.81 0.420 -1357.054 3248.593 _cons | -9733.014 34579.06 -0.28 0.779 -77732.2 58266.17

20

Conclusions For the linear model considered here, the coefficient measures the marginal contribution of the independent variable to the dependent variable, holding all other variables fixed. In our case corruption proved to be positively correlated with the dependent variable in one case.

Yet the standard error of the obtained coefficients seem to estimate a very large amount of noise in the coefficient estimates.

The p-values, always being, above 5% enforce us to reject the null hypothesis of a zero coefficient.

In terms of the R-squared statistic which we interpret as the fraction of the variance of the dependent variable explained by the independent variables we conclude that adding a variable for corruption does not imply more explaining power into the equation. The value of this statistic is almost identical in cases of regressions with and without the Kraay Index included.

The above result show, that the measured level of corruption does not affect the ISO or export flows in either direction: DC-LDC nor DC-DC. Adding such a variable does not seem to give any additional explanatory value to the equation. This might be the result of the fact, that measured corruption may be both a pull or push factor for businessmen generating capital and goods’ flows into an other country.

An other issue is that the measure of corruption would rather be a purely academic feature which is not necessarily taken into consideration when starting a certain form of economic cooperation between countries.

21

Bibliography Habib, Mohsin and Leon Zurawicki, Corruption and Foreign Direct Investment, Journalof International Business Studies, Vol. 33, No. 2. (2nd Qtr., 2002), pp. 291-307.

Kaufmann, Danial, Aart Kraay, and Massimo Mastruzzi, Governance Matters, V: Aggregate and Individual Governance Indicators for 1996–2005, World Bank Policy Research Working Paper 4012, September 2006.

Shang-Jin Wei and Andrei Shleifer, Local Corruption and Global Capital Flows, Brookings Papers on Economic Activity, Vol. 2000, No. 2. (2000), pp. 303-354.

Shannon W. Anderson, J. Daniel Daly, Marilyn F. Johnson, Why Firms Seek IS0 9000 Certification: Regulatory Compliance Or Competitive Advantage, University of Michigan Business School, Ann Arbor, Michigan, USA, Boston College, Carroll School of Management, Chestnut Hill, Massachusetts USA

22

Bibliography

1. Anderson, Daly, Johnson; Why firms seek IS0 9000 certification: regulatory compliance or competitive advantage?, University of Michigan Business School, Ann Arbor, Michigan, USA, Boston College, Carroll School of Management, Chestnut Hill, Massachusetts USA 1999 2.

23