Comments and Questions

Comment on the Discussion Paper 2008-6, March 3, 2008 “The Economic Effects of Croatia’s Accession to the EU” by Arjan Lejour, Andrea Mervar and Gerard Verweij If you have a model – a big one – you have to apply it as often as possible in order to recoup the costs of constructing such a model. By doing so, it makes no difference whether the topic you address is suitable for it or not. In this case the WorldScan model of the Netherlands Bureau for Economic Policy Analysis – formerly called Central Planning Bureau (CPB) - is used to assess the economic effects of Croatia’s accession to the EU. This seems to me a classical case of “shooting with cannons on sparrows”. Some of the authors, together with other co-authors made similar exercises with WorldScan in the cases of the big 2004 EU enlargement and of Turkey’s ambition to become a member of the EU. Now Croatia is on the agenda to replicate these exercises in evaluating the economic outcome of its EU accession. The results of the WorldScan model simulations are accordingly: the mini state Croatia with a population twice that of Vienna which produces a GDP of ½ percentage point of that of EU27 wins by acceding the EU, whereas this event has no economic impact at all in the EU. This result is plausible, but one would have reached it simply by drawing on the common sense. After this somewhat exaggerated introductory remarks let me go further into detail. Of course, a computable general equilibrium model like the WorldScan has advantages. One can have a look not only on macro-economic variables but also on sectoral ones and such a model is a consistent replication of modern economic theory – although, which I will comment on later – with some caveats. For this purpose the WorldScan model was aggregated such that it captures 15 sectors and a large number of countries and world regions, in particular Croatia and the EU, divided into six regions. The database relies on that of GTAP of the year 2001. First, the authors briefly describe the Croatian economy, its structure and its development since it declared its independence in 1991 after the break-down of Yugoslavia. Then they sketch the major integration steps of the Croatian government to approach the EU – which is valuable information for those not familiar with Croatia. Then they construct a baseline scenario up to 2025. This scenario implies an extrapolation of the status quo with some plausible assumptions on economic growth (higher that that of the EU) and no reforms in Croatian economic policy. By comparing the integration scenario (representing the participation of Croatia in EU’s internal market) the study deduces the economic effects of Croatia’s accession to the EU. The question is now how the authors model the accession to the internal market. This is done by just two scenarios – both of which operate via trade effects:(i) With the accession Croatia – like any other newcomer – takes part in the customs union of the EU, implying the take-over of the common external tariff (CET) and the participation in the common external trade policy of the EU; further more taking part in the internal market implies the elimination of border controls; the former is modeled by adjusting the still existing tariffs of Croatia to those of the CET in all sectors and the latter is modeled by a tariffication of NTBs (or trade costs). The trade advantage of becoming a member of the EU is estimated by gravity equations for each of the 15 sectors with an EU dummy. In 12 out of 15 sectors the dummy has the correct (positive) sign. The resulting potential trade increase by sector due to the dummy effect is translated into a Samuelson iceberg trade-cost equivalent (NTB). The EU dummy is estimated out of a sample of 38 countries which probably have completely different bilateral trade relations than the small country Croatia which – as a rule for a small open economy – trade primarily with neighbouring countries.(ii) It is assumed that EU membership will stimulate institutional reforms in Croatia. This effect is simulated by re-estimating the gravity equations for bilateral trade with the Perception Corruption Index (PCI) published by Transparency International. According to the authors’ estimation the CPI enters significantly into the gravity equations. An improvement stimulates exports! Although the endeavour to capture the trade effects is admirable, EU membership is much more complex than just considering the trade creation effects of the internal market:(i) Participating in the Internal Market (IM) of the EU not only implies the working of one freedom (free trade), but also those of the other three “F’s”: freedom of offering services, freedom of movement of labour and capital. The model only captures one “F” – free trade. Furthermore the Internal (or Single) Market implies more competition and a compliance with the common competition policy. The WorldScan model, however is not able to capture market imperfections and economies of scale. It is also not able to capture dynamic effects of the IM, long ago postulated by Baldwin. Also the possible growth-stimulating impulses by investing the receipts Croatia – as a poor country - will get out of the structural funds of the EU are not considered in the simulations (see page 16 in the paper). The simulations with WorldScan only capture the static allocative efficiency gains from EU accession (see page 18 in the paper) .(ii) The authors take a rather optimistic view (those of the Croatian government?) by assuming that Croatia will become EU member in 2009.(iii) On top of the participation in the IM one can expect that Croatia will introduce the Euro and hence participate in EMU relatively early – probably before the end of the simulation horizon, 2025. But the effects of the introduction of the Euro are not taken into consideration in the simulations. I guess that the WorldScan model is not suitable to deal with such an integration step. The macroeconomic results of participating in the IM are plausible: Croatia’s GDP will increase by 1.1 percentage points in 2025 . This implies an annual GDP gain of 0.06 percent. More plausible than the GDP effects are the welfare effects. Liberalization of trade increases welfare. But does liberalization of trade necessarily lead to GDP growth? In Table 4.1 the authors only report the increase of Croatia’s export volume (by 13.9%) and not the development of the import volume (this is done in the text on page 20); accordingly they increase by 15.9% and hence much faster than Croatia’s export which implies an increase of the deficit in its trade balance. Simple national accounting would hence imply not a positive effect on real GDP but a negative one. Where, then does the growth effect come from? On page 19 the authors try to explain the channels which lead to GDP growth: (i) one is the exploitation of comparative advantages (?) and (ii) a positive terms of trade effect. The experience so far with the most recent EU enlargements (2004 and 2007) shows that the old EU member states have exerted a much higher competitive pressure than the new ones. The consequence is a deterioration of the trade balances in the new member states. A similar effect was detected in the case of Spain’s EU accession in 1986. Why should this be different with Croatia?Although the sectoral impact of the EU accession of Croatia would be interesting and informative, I refrain from commenting the results of Table 4.2, simply because they depend on the estimated parameters of the gravity equation, which refer to an economic structure far back in the past. In the near future – up to 2025 – the composition and structure of the Croatian economy will change in the ongoing transformation process and will therefore not be the same as it was in the base year of the data set, 2001. Furthermore, most of the (trade) integration effects have already been consumed before Croatia will enter the EU. This is due to the asymmetric trade liberalization via the trade agreements with the EU in the Stability and Association Agreement (SAA) – similarly to the Europe Agreements with the CEECs before they entered the EU in 2004 and 2007! Such pre-accession integration effects are captured in the case of Bulgaria’s and Romania’s accession to the EU in the simulations by Breuss (2007).The effects of the “New CEFTA” arrangement is not mentioned explicitly, only indirectly and vaguely when the authors (on page 7) speak about the fact that “Croatia has arrangements on free trade zones with a number of neighbouring countries”! Nothing is said explicitly about the implications for the direct neighbours of Croatia – in the Western Balkans – and the indirect neighbours - Austria, Hungary and Italy. The WorldScan simulation evaluates the integration (i.e. the trade effects) of Croatia’s accession to the EU only at the end of the simulation period, in 2025. Unfortunately, the model does provide no information on the transitional path from the EU accession 2009 (or later) until 2025. The macroeconomic (and more so the sectoral) results of institutional reforms in Croatia after becoming a member of the EU are much more problematic than the trade effects of entering the IM. By assuming that Croatia will be able to reduce its level of corruption to that of Portugal, its welfare would increase by 5.6 percentage points and real GDP in 2025 would be higher by 7.8 percentage points. These very mystic effects – simulated with a neoclassical CGE model – lead to implausible high economic effects. That less corruption will increase trade so dramatically may be captured by respective dummy variables, nevertheless it is “bad political economy”! Résumé:1) It seems that WorldScan is to big a model to capture the essential effects of integrating a small country into the EU. 2) Although WorldScan is quite sophisticated in modeling the bilateral trade flows it is not able to capture the essential effects of a country participating in the EU Internal Market. 3) The study captures only trade effects and assumes that more trade leads to more growth, an assertion which is over and over again questioned in the literature on “liberalization of trade and growth”. 4) The effects of institutional reforms on GDP growth (via the trade channel) are completely “fictitious”. Even if one would believe such channels one must only look at the (bad) examples of the most recent EU enlargement – Bulgaria and Romania – to learn that EU membership is not necessarily connected with enough pressure to reduce corruption and induce real reforms. 5) I believe that EU membership may contribute – at least in the long-run – to more GDP growth than simulated in this study but due to completely other channels than just trade effects. Primarily more growth is stimulated by fiercer competition in the internal market and hence also by an increase in productivity. The latter may also be stimulated by participating more intensively in EU’s Research Framework Programmes. Some minor technical remarks which could help to improve the paper:1) In the Abstract, third line: at the beginning of the sentence “Consumption per capita …” one should write “C” in capital letters.2) Table 2.1: it is not clear whether “GDP (billions €) is in PPs or at actual exchange rates;3) Tabel 2.3: the headlines of the third and fourth column are not clearly specified;4) At the end of page 8, CBS is quoted as 2006 (in the list of references one finds CBS, 2006a and 2006b – which one is the correct quotation?5) Place the number of equation (1) at the right side of the page (on page 11).6) Table 4.1 and Table 4.3: Why no also report the figures for “Import volume”?7) Eurostat (2006) is mentioned in the footnote to Table 2.1, but not in the list of references.8) The title “European Commission, 2006b” is in the list of references but not detectable in the text; the same is true for the title “Nahuis, 2004”. References:Breuss, F., A Prototype Model of EU´s 2007 Enlargement, Europainstitut at the Vienna University of Economics and Business Administration, EI Working Papers, No. 76, Vienna May 2007 (downloadable under: http://www.wu-wien.ac.at/wuw/institute/europainstitut/pub/workingpaper/wp76).An updated and revised version can be found by the same author and with the same title under: FIW Working Paper Series, No. 7, July 2007 (downloadable under: http://www.fiw.ac.at/fileadmin/Documents/Publikationen/Working_Paper/N_007-breuss.pdf)