Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Trade Liberalization, Financial Development and Economic Growth in The Long Term: The Case of Turkey

Trade Liberalization, Financial Development and Economic Growth in The Long Term: The Case of Turkey T he determinants of economic growth have been a much debated theoretical issue in the literature, especially after the endogenous growth theory of the late 1 980s. T his new theory highlights the importance of economic policies that lead to an increasing rate of return. In particular, it is argued that human capital, trade liberalization and financial development may play very important roles in the determination of economic growth. T his paper tries to empirically estimate the joint impacts of trade liberalization and financial development on economic growth for the period 1 9602 004. Instead of using common proxies for the issue, principal components analysis is employed to develop better measures (indexes) for trade liberalization, financial development and the joint effects of both. T he empirical results obtained from the J ohansen co-integration procedure show that trade liberalization, financial development and the joint impacts of both positively contributed to economic growth in T urkey for the period 1 963-2 005. Keywords: Trade liberalization, Financial development, Economic growth, Co-integration, Endogenous growth theory, Turkey. J EL: C32 , E44, F 1 3, O 1 1 , O 1 6. DOI: 10.2478/v10033-008-0012-x 1. Introduction The im pact of financial and trade regim es on econom ic growth have been a hotly debated theoretical issue, particularly after the em ergence of the endogenous (new) growth theory during the last two decades. Contrary to the neo-classical growth theory, the new growth theory im plies that the econom ic policies followed in a country m ay have a significant im pact on the long-term rate of econom ic growth. To this end, m ost developing countries that form erly followed restrictive econom ic policies have started liberalizing their trade and financial sectors in order to increase econom ic growth in the 1980s. The m ain argum ent for this policy change was that both trade and financial liberalization policies reduce inefficiency in the production process and positively influence econom ic growth. Following this line of reasoning, Turkey, as a developing econom y, has witnessed an unprecedented and staged reform attem pt involving external (trade) and internal (financial) liberalization, especially after the 24 January Decisions following the econom ic crisis in 1980. In Turkey, econom ic liberalization in term s of trade and financial sector was at the heart of the stabilization program m e Novem ber 2008 *Kar Faculty of Economics and Administrative Sciences, Kahramanmara Sütçü mam University, Kahramanmara/TURKEY e-mail: mkar@ksu.edu.tr *Peker Faculty of Nazilli Economics and Administrative Sciences Adnan Menderes University, Aydin/TURKEY e-mail: ottopeker@yahoo.com *Kaplan Faculty of Economics and Administrative Sciences, Nide University, Nide/TURKEY e-mail: mkaplan@nigde.edu.tr em ployed in 1980 and constituted an integral part of the econom ic policies since then. The theoretical core of the econom ic reform program m e is to provide efficiency in the allocation of scarce resources. In the em pirical literature, the im pact of trade liberalization 1 and financial developm ent2 on econom ic growth has been separately exam ined for Turkey. However, there has been theoretical literature that highlights the joint im pact of these issues on econom ic growth. The joint im pact of both variables is initially highlighted in the Roubini and Sala-i Martin (1991) m odel, which has extended the Barro (1991) growth m odel by incorporating both factors. The inclusion of both trade and financial variables in the Barro (1991) m odel showed that financial and trade variables m ay play a very im portant role in econom ic growth. Recently, Blackburn and Hung (1998) have presented a theoretical analysis of the joint im pact of both financial developm ent and trade liberalisation on econom ic growth. Using endogenous growth theory, their m odel predicts that econom ic growth rates in the presence of financial interm ediation tend to be higher than those under direct lending and borrowing. The m odel also predicts that both financial developm ent and trade liberalization jointly facilitate the rate of econom ic growth by decreasing redundant research efforts and increasing m arkets for new products. Therefore, this paper em pirically exam ines the im pacts of trade liberalization and financial liberalization and/ or developm ent on econom ic growth in Turkey by using a sim ple endogenous growth fram ework and by m aking use of new developm ents in tim e series techniques for the period 1963-2005. This paper is distinguished from the earlier existing literature in two aspects. First of all, this paper tries to assess the joint im pact of trade liberalization and financial developm ent on econom ic growth. Second, there are different m easures for trade liberalization and financial developm ent in the literature and the existing studies em ploy only one of these proxies in their analysis. However, the trade and financial liberalization affect econom ic growth through different channels and each proxy captures a single aspect of the issue. In order to overcom e this problem , three com posite indexes, for trade liberalization, for financial developm ent 1 The relationship between growth and trade liberalization is usually exam ined in the em pirical literature in two different lines for Turkey: The first line of the existing em pirical research tries to assess the im pact of openness on econom ic growth (Conway, 1987; Greenaway and Sapsford, 1994, 1995; Ghatak, Milner and Utkulu, 1995; Subasat, 2002; Utkulu and Ozdem ir, 2005, Hilm i and Safa, 2007). The second line of the em pirical works exam ines the relationship between trade liberalization and the total factor productivity (Krueger and Tuncer, 1980, 1982; Nishim izu and Robinson, 1986; Foroutan, 1991; Ozm ucur and Karatas, 1994; Filiztekin 2000). 2 With the em ergence of financial liberalization hypothesis, the em pirical studies tried to assess the im pact of financial liberalization on econom ic growth by sim ple regressing growth rate on the real interest rate (Fry, 1978, 1979, 1980). Furtherm ore, the relationship between financial developm ent and econom ic growth is analyzed in term s of causality issue in a bivariate m odel (Kar and Pentecost, 2000; Unalm is, 2002; Ozatay and Sak, 2002). In addition, Yeldan (1997), Lewis (1992) and Karabulut and Dem iroz (2002) exam ined the im pact of financial liberalization/developm ent on econom ic growth in the fram ework of the CGE m odels. Finally, Guncavdi and Kucukciftci (2005) investigate the role of financial reform s on econom ic growth with a m ethodology based on the Leontief's input-output m odel and conclude that the production sector of the econom y has increasingly becom e independent from the use of financial services produced by the banking system in the post-reform period. and for a narrow sense econom ic liberalization, are constructed by applying principal com ponents analysis. The structure of this paper is as follows: the next section presents the theoretical linkages am ong trade liberalization, financial developm ent and econom ic, growth and introduces the aggregate production function to be estim ated. Section three gives brief inform ation about the Turkish econom y, particularly highlighting the developm ents in trade and financial sector. In section four, the m easures for trade liberalization and financial developm ent are reviewed and three indexes are constructed by em ploying principal com ponents analysis. In section five, the em pirical results are presented using econom etric techniques and the outcom e of the longterm production function is interpreted in detail. The paper finishes with a conclusion. 2.Theoretical Framework: Trade Liberalization, Financial Development and Economic Growth With the em ergence of the endogenous growth theories in 1980s, the relationship between econom ic policy and growth becam e a highly debated issue. In the theoretical literature, discussions are focused on different channels through which econom ic policy affects econom ic growth. In this section, different m odels of growth will be discussed to provide a fram ework of thought that helps to understand the im pacts of each link between policy and growth. In other words, each channel through which econom ic policy affects growth has different im plications for growth in different m odels. There are two com peting theoretical fram eworks in the growth literature, nam ely neo-classical and endogenous growth theories. The m ain differences between them are whether the policy change has a longterm effect on the growth rate. On the one hand, the neoclassical theory argues that a policy change has no effect on long-term econom ic growth and, on the other hand, that endogenous growth theory shows a policy change in econom y does m atter. The endogenous growth theory is a reaction to the traditional Neo-Classical growth m odels3 , represented by, am ong others, Solow (1956). This new approach to growth theory has sought to supply the m issing explanation of longterm growth. In essence, this approach provides a theory of technical progress, one of the central m issing elem ents of the neo-classical m odel. In other words, endogenous growth theories seek to discover what lies behind the exogenous rate of technical progress and hence a country's growth rate. Endogenous growth theory recognizes that technological change occurs as a result of the efforts of profit-m axim izing firm s to invent new blueprints, and that technological progress is an endogenous outcom e of econom ic activity. 3 For a recent em pirical and theoretical review of the literature on growth, see Renelt (1991), Herm es (1994), Levine (1997, 2001), Thirlwall (2000), Favara (2003), Auerbach and Siddiki (2004), Subasat (2002), Winters (2004). SEE Journal The crucial distinction between `old' and `new' growth theories is that the form er utilizes the assum ption that returns to the capital stock is dim inishing, while the latter argues that returns to capital itself or, in a wider sense, to the stock of physical and hum an capital form ation is constant or increasing (Sala-i Martin, 1990a). This then im plies that those variables that lead to non-decreasing returns drive the growth rate. Num erous candidates have been recom m ended as the source of non-decreasing returns: particularly, the stock of hum an capital Lucas (1988); accum ulated capital, Rebelo (1991); research and developm ent, Rom er (1986, 1990); or public infrastructure investm ent (Barro, 1991). Thus, endogenous growth m odels highlight sectors of the econom y that influence the growth path of an econom y. This can be sim ply shown in a Robelo-type production function, known as the AK m odel. Most of the endogenous growth m odels can be viewed as extensions or m icro-foundations of the AK m odel (Sala-iMartin, 1990b). Rebelo (1991) form ulated the sim ple form of the endogenous growth m odel, which has since been widely used in em pirical analysis. The AK m odel takes its nam e from its production function. In its original form , the m odel setting involves dynam ic m axim ization. In this section, we will m ake the further assum ption of a constant savings rate. This assum ption, however, does not change the m ain conclusions and intuitions of the m odel. In the AK m odel, the production function takes the following form : the cap ital accum ulates. T his can b e shown easily: after sub stituting the value of investm ent into eq uation (3) and then d ivid ing b oth sid es b y K t -1 and taking the logarithm of b oth sid es, the resulting eq uation will b e: K log t = log[sA +(1 - )] K t -1 F or sm all values of s, A and (4) can written as: and (4) sA > , eq uation log K t = sA - (5) T his eq uation says that the rate of growth of cap ital stock is constant if tariff rates are constant. A fter taking the logarithm and d erivative of the p rod uction function and sub stituting the value of the eq uation of m otion of the cap ital from eq uation (5) and the value of return to cap ital from the eq uation (2), the longterm rate of growth of outp ut can b e written as follows: log Yt = log K t = s -s (6) Yt =A K t (1) where Yt represents output, K t is capital stock at tim e t and A is som e positive constant. This form ulation of the production function m eans that there are constant returns to capital accum ulation. It is also im portant to note that A is equal to the return to investm ent in this m odel. As will be explained in the next section, trade policy prim arily affects the rate of return of capital and hence growth. Therefore, A can be written as a function of trade policy () as, A= (2) Equation (2) indicates that the rate of return of capital is a negative function of trade policy. The accum ulation of capital is form ulated as: K t = I t -1 + (1 - ) K t -1 and I t = sYt (3) where s is the investm ent rate and is the depreciation rate. Both are assum ed constants, and investm ent at tim e t ( I t ) is equal to the savings in the econom y. The special form ulation of the production function in the AK m odel (equation 1) im plies that the m arginal product of each unit of capital is always equal to A. It does not decline as Novem ber 2008 F rom eq uation (6), it is ob vious that the rate of growth of the econom y is d ecreasing with tariff rates and increasing with saving rates. H ence, any econom ic p olicy that increases the return to investm ent will p erm anently increase the rate of growth of the econom y. A lm ost allend ogenous growth literature has concentrated on the d eterm inants of the return to investm ent, A , and how p olicy change affects it (S ala-i-M artin, 1 990a). E q uation (6) has two m ain im p lications in term s of econom ic p olicy change. E conom ic p olicies in the trad e and financial sectors will have a long-term effect on econom ic growth. W hile trad e p olicy affects the econom ic growth through the change on tariff rates, the saving rates are influenced b y a well-functioning financial system . T he theoretical linkages am ong trad e lib eralization and financial lib eralization and econom ic growth can b e exp lained as follows. T rad e p olicy in term s of tariff red uction or elim ination of restrictions on trad e m ight have im p acts on the growth through several channels. If op enness is to affect econom ic p erform ance, it m ust have an interm ed iate effect on one or all of the following: (i) allocation of factors of p rod uction across sectors - the allocation effect (Y oung, 1 991 ; R ed d ing, 1 997 ; G rossm an and H elp m an, 1 992); (ii) op enness will increase com p etition in the d om estic econom y and hence p rod uctivity - the im p ort d iscip line hyp othesis (G reenaway and M ilner, 1 993; A ghion, D ewatrip ont and R ey, 1 997 ; A ghion, H arris and V ickers, 1 997 ; A ghion and H owitt, 1 996); (iii) op enness enlarges the m arket for d om estic p rod ucers, which they can take ad vantage of - the scale econom ies (T aylor, 1 994; G rossm an and H elp m an, 1 991 ); (iv) op enness increases the num b er of inp uts that have no d om estic sub stitutes and thus lead s to a higher cap acity for utilization and p rod uctivity - the availab ility of inp uts (N ishim izu and R ob inson, 1 986; Q uah and R auch, 1 990; R ivera-B atiz, and R om er, 1 991 ; G rossm an and H elp m an, 1 992); and , finally, (v) the flow of knowled ge across sectors and countries the sp illover effect (F ed er, 1 982; G rossm an and H elp m an, 1 992). O n the other hand , the role of financial sector in econom ic d evelop m ent has long b een one of the hotly d eb ated issues am ong econom ists (S chum p eter, 1 91 1 ; G old sm ith, 1 969; P atrick, 1 966; M cK innon, 1 97 3; S haw, 1 97 3). W ith the em ergence of the end ogenous growth theory, several stud ies have attem p ted to show how the op eration and p olicies of the financial sector m ay affect the rate of econom ic growth (G reenwood and Jovanovic, 1 990; B encivenga and S m ith, 1 991 ; K ing and L evine, 1 993a, 1 993b ; R oub ini and S ala-i M artin, 1 992; P agano, 1 993; L eigh, 1 996; D em etriad es and H ussein, 1 996; A restis, 2005; S id d iki, 2002; A uerb ach and S id d iki, 2004; A restis, 2005; L iang and T eng, 2006). F inance can influence growth in an end ogenous growth m od el through increasing the savings rate (B encivenga and S m ith, 1 991 ), b y increasing the returns on investm ent (G reenwood and Jovanovic, 1 990), and b y increasing hum an cap ital accum ulation. F rom a m acroeconom ic or aggregate p rod uction function p oint of view, all this m eans that econom ies that are d evelop ed m ore financially will b e ab le to transform a given am ount of inp uts, K , into larger am ount of outp ut, Y . T his is why the p rod uction function is an increasing function of the financial d evelop m ent of the econom y (R oub ini and S ala-I M artin, 1 992). term econom ic d evelop m ent p lan. T his p rop osal was for three five-year-p lans, the first of which b egan in 1 963. T he econom ic d evelop m ent p lans were aim ed at p rod ucing a well-b alanced econom y with p rogress in agriculture and ind ustry, thereb y m aking the m ost of the availab le resources in the T urkish econom y. T he m ain features of this p eriod are that the econom ic p olicies carried out within the d evelop m ent p lans were characterized as interventionist and p rotectionist. A ccord ingly, p olicies were m ainly d esigned to p rotect d om estic ind ustry from foreign com p etition and to increase the governm ent control over the allocation of resources and p rod uction of good s. T hese econom ic p olicies im p lem ented in the early 1 960s were p ursued for ab out twenty years, until 1 980, when the T urkish econom y found itself in a m ajor crisis. In 1 980, a new stab ilization p olicy was accep ted and the etatism and im p ort sub stitution p olicies were switched to an exp ort-oriented ind ustrialization p olicy b ased on a m arket m echanism . T he p olicy p ackage p ut into effect in 1 980 and reinforced in the following years was m ore than just a stab ilization and ad justm ent p ackage; it also m arked a shift in d evelop m ent strategy from inward orientation to outward orientation (Y ild izoglu and M argulies, 1 988; S enses, 1 984; O nis, 1 986; D ervis and P etri, 1 987 ). T rad e and financial lib eralization were the m ain p olicy tools in this stab ilization p rogram m e. 3.2. Trade Liberalization in Turkey A fter p ursuing the "im p ort sub stituting ind ustrialisation strategy (IS I)" as a d om inant ind ustrialization strategy in the 1 950s, 1 960s, and 1 97 0s, T urkey switched to an outward -oriented ind ustrialization strategy with the IM F sup p orted stab ilization p rogram m e that was introd uced to resum e growth following the econom ic crisis in 1 980. R ap id exp ort growth was one of the m ain ob jectives of the 1 980 stab ilization p rogram m e to im p rove the huge trad e d eficit, restore international cred itworthiness and estab lish the cred ib ility of lib eralization reform s at hom e. A variety of incentives were introd uced to p rom ote m anufactured exp orts. T hese incentives includ ed tax reb ates, cred it sub sid ies, and foreign exchange allocated for the im p ort of interm ed iate p rod ucts. T he success of the lib eralization p rocess in the 1 980s p rom p ted the governm ent to p ursue further lib eralization in the 1 990s. T herefore, T urkey lib eralized her im p ort regim e b y ab olishing the d ep osit req uirem ent for im p orts and the im p ort licensing system in the early 1 990s. A ccom p anying agreem ents with the W orld T rad e O rganisation (W T O ) in 1 994 have significantly contrib uted to the lib eralization of the im p ort regim e. A s a m em b er of the W T O , T urkey has ad op ted the rules and p roced ures governing the m ultilateral trad ing system and entered into negotiations with several E astern and C entral E urop ean, M ed iterranean and B altic countries to conclud e free trad e agreem ents. 3. A Brief Review of Turkish Econom y 3.1. Macroeconom ic D evelopm ents It is very com m on to exam ine m acroeconom ic d evelop m ent in the T urkish econom y und er three sub -p eriod s, 1 923-1 960, 1 960-1 980, 1 980 and after, d istinguished not only b y d ifferent structural cond itions, b ut also b y the governm ent's resp onse to those cond itions. T here was heavy state intervention form ulated as "etatism " b efore 1 960 (O kyar, 1 965). E tatism , an econom ic p olicy excessively controlled b y the S tate, b ecam e unp op ular and heavily criticized am ong T urkish acad em ics and b usinessm en. W ith the changes in the p olitical structure in term s of the m ulti-p arty system , there was a p olicy change toward s lib eralization in the p eriod 1 950-53. H owever, with the d eterioration of m acroeconom ic structure, the etatist econom ic p olicies were re-initiated after attem p ts at p olicy change. A fter the 1 960 m ilitary coup , T urkey entered into a p eriod of d evelop m ent p lans. In S ep tem b er 1 960, the S tate P lanning O rganisation (S P O ) was estab lished to stud y the T urkish econom y and to p rop ose and enforce a long28 In 1 996, T urkey entered into a new era b y signing a C ustom s U nion A greem ent with the E urop ean U nion (E U ). A fter the lib eralization p rogram m e in 1 980, this was the second m ost im p ortant d evelop m ent affecting the T urkish econom y as a whole. E xcep t for sensitive p rod ucts, m ainly m otor vehicles, footwear, and furniture, T urkey lifted all tariff and non-tariff b arriers for m anufacturing p rod ucts originating from the E U . T urkey also ad op ted the E U 's C om m on E xternal T ariff for good s im p orted into T urkey from third countries. T his req uired a further lib eralization of her tariff regim e, since T urkish p rotection rates were higher overall than the C om m on E xternal T ariff (H arrison et al., 1 996; T ogan, 1 997 ). M oreover, T urkey had to red uce her tariffs to countries that signed a P referential T rad e A greem ent (P T A ) with the E U . 3.3. Financial Liberalization in Turkey P rior to 1 980, T urkey was a typ ical exam p le of highly restricted and segm ented financial m arkets (A kyuz, 1 990). Interest rates were d eterm ined institutionally and kep t at artificially low levels. B y the end of the 1 97 0s, real interest rates b ecam e highly negative d ue to the acceleration of inflation (F ry, 1 97 9). T he state owned b anks were d om inant institutions in the T urkish financial system (F ry, 1 97 9). E ntry into the b anking sector (d om estic and foreign) was restricted . S ince then, T urkey has b een exp eriencing a lib eral ap p roach to its financial m arkets as a key com p onent of the newly ad op ted growth-oriented structural ad justm ent p rogram since 1 980 (A ricanli and R od rik, 1 990). W hen the authorities lifted the ceilings on p ersonal tim e d ep osit rates and lend ing rates were ab olished 4 . A t the tim e it was consid ered a "m ajor step in d eregulation of interest rates which b reaks a p ractice that has b een in force som e 50 years" (W olff, 1 987 :1 04). T his p olicy change was not very sm ooth and the reluctant b ehaviour of the financial institutions req uired the authorities to intervene into the sector two years after the im p lem entation of financial lib eralization and to d eterm ine interest rate for a p eriod . 5 A nother d evelop m ent, in the first half of the 1 980s, was that resid ents (and non-resid ents) were allowed to op en foreign exchange d ep osits in com m ercial b anks (R ittenb erg, 1 988). T he essential regulation was finally initiated in 1 985, when the new B anking L aw was enacted . T he law introd uced new regulations in term s of p rovision for a m inim um cap ital and a cap ital ad eq uacy ratio. T he ownership structure of b anks was also regulated . F urtherm ore, a B ank S up ervision unit at the C entral B ank b ecam e op erational in 1 986. T hus, five years after the initial lib eralization of d om estic interest rates, an ad eq uate regulatory and institutional fram ework 4 T he d evelop m ents in financial sector since 1 980 are exp lained in m ore d etail A kyuz (1 990), C osan and E rsel (1 986), Inselb ag and G ultekin (1 988), C izre-S akallioglu and Y eld an (2000), O zatay and S ak (2002) and R ittenb erg (1 988). 5 A tiyas (1 990) exp lains in d etail of the resp onse of p rivate sector to financial lib eralisation. was d efined and b ecam e op erational. A fter the lib eralization of the cap ital account in 1 989 and estab lishm ent of the sup ervisory and regulatory unit at the C entral B ank, it was b elieved that an ad eq uate regulatory and institutional fram ework was finally d efined and b ecam e functional. H owever, the crises of N ovem b er 2000 and F eb ruary 2001 have shown that these d evelop m ents were not sufficient to have stab le, efficient and wellestab lished financial m arkets in T urkey. T hese crises led not only to the estab lishm ent of a new institution, nam ely the S up ervisory and R egulatory B oard of B anking , which aim ed at restructuring the financial system in 2000, b ut also to the d evelop m ent of new p olicies that p rop osed an efficient and effective financial m arket. It can b e argued that the T urkish financial system finally had an ind ep end ent sup ervisory and regulatory b od y, 20 years after b eginning to im p lem ent lib eralization p olicies in the financial sector. 4. Measurem ent of Trade Liberalization and Financial D evelopm ent 4.1. Measurem ent of Trade Liberalization R esearchers in the recent em p irical literature concentrate on find ing reliab le p roxies of trad e lib eralization. H owever, the share of exp ort as a p ercentage of incom e, the share of the im p ort as a p ercentage of incom e, the share of exp ort p lus im p orts (trad e volum e) as a p ercentage of the incom e and tariffs constitute very com m on p roxies for trad e lib eralization in the em p irical literature. In this article, the following p roxies of trad e lib eralization are em p loyed in the em p irical analysis. E xp ort to G D P ratio (X /G D P ): T he first theoretical channel that links op enness to econom ic p erform ance goes through the allocation of resources. A ccord ing to this argum ent, op ening up to international trad e b rings ab out reallocation of resources accord ing to com p arative ad vantages (G rossm an and H elp m an, 1 992, Y oung, 1 991 ). S ince the d irect effect of the allocation of resources is ob served on the level of exp orts, the share of exp orts in total p rod uction can b e used to rep resent this d im ension of op enness. In ad d ition, the share of exp orts in p rod uction can b e used as a p roxy of op enness to cap ture the d im ension of op enness related to scale econom ies and the availab ility of inp uts. Im p ort to G D P ratio (M /G D P ): T he im p ort share in total p rod uction can b e used as an op enness p roxy characterizing the d im ension of op enness related to increased international com p etition. It also rep resents the allocation effect of op enness since the im p orts of those sectors that have com p arative d isad vantages will increase following trad e lib eralization. F oreign trad e to G D P ratio (X + M /G D P ): T he share of the total of exp orts and im p orts in total p rod uction p rovid e the p roxy that rep resents the technology sp illover d im ension of op enness. O p enness to trad e facilitates access to the technological inform ation in the world (G rossm an and H elp m an, 1 992, C hp -9), which creates technological sp illover through im p orts as well as exp orts. 4.2. Measurem ent of Financial D evelopm ent O ne of the m ost d ifficult asp ects of em p irically investigating the relationship b etween financial d evelop m ent and econom ic growth is the m easurem ent of "financial d evelop m ent". H owever, the p ractitioners are forced to form a well-d efined set of m easures of financial d evelop m ent b y the availab ility of d ata at hand . T he p roxies p rop osed for m easuring the level of financial d evelop m ent are b asically chosen from the m onetary and cred it aggregates in an econom y. T he rationale for the inclusion of a wid e range of p roxies is to m axim ize the inform ation on financial d evelop m ent. In other word s, d iverse aggregates should b e ab le to catch d ifferent functions of the financial m arkets. In this article, the following p roxies6 for financial d evelop m ent are em p loyed in the em p irical analysis. N arrow M oney R atio (M 1 /Y ): In the ab sence of the financial sector econom ic agents have to hold their financial assets at hand . T his m eans that the fund s for investm ent will b e kep t out of the financial sector in the econom y. W ith financial d evelop m ent, the ratio of narrow m oney to incom e will d ecrease. N arrow M oney B road M oney R atio (M 1 /M 2): In a fragm ented financial sector, econom ic units m ay p refer to hold their fund s out of the financial sector to rem ain liq uid . H owever, d evelop m ents in the financial system in term s of not only organizations such as b anks b ut also instrum ents in this sector m ay lead p eop le to p ut their m oney in the b anking sector, through which investm ent can b e carried out. In short, with financial d evelop m ent, d ep osits in the b anking sector m ay b e increased and as a result the M 1 / M 2 ratio will d ecrease. B road M oney R atio (M 2/Y ): M onetary aggregates also p rovid e an alternative set of variab les to m easure the extent of financial d evelop m ent (D e G regorio and G uid otti, 1 995; G aletovic, 1 996; L ynch, 1 996). In the literature, the com m only used m easure of financial d evelop m ent is a ratio of som e b road m easure of the m oney stock, usually M 2, to the level of nom inal incom e (G elb , 1 989; K ing and L evine, 1 993a, 1 993b ; M urind e and E ng , 1 994a, and 1 994; L yons and M urind e, 1 994; D em etriad es and H ussein, 1 996; A restis and D em etriad es, 1 997 ; K wan et. al., 1 998). T his sim p le ind icator m easures the d egree of m onetization in the econom y. T he m onetization variab le is d esigned to show the real size of the financial sector of a growing econom y. M oney p rovid es valuab le p aym ent and saving services. T he `narrow m oney' stock b est reflects the form er 6 F or a m ore com p rehensive set of m easures of financial d evelop m ent, see L ynch (1 996). and `b road m oney' the latter. N arrow m oney b alances should rise in line with econom ic transactions, b ut b road m oney should rise at a faster p ace, if financial d eep ening is occurring (L ynch, 1 996). It is argued that the use of m onetary aggregates as a p roxy for the d egree of financial d evelop m ent m ight also p resents p rob lem s (D e G regorio and G uid otti, 1 995; K ing and L evine, 1 993a). K ing and L evine (1 993a) note that d ifferent d efinitions of m onetary aggregates m ay act as p roxies for d ifferent roles of financial interm ed iation. In som e cases m onetary aggregates m ay b e very p oor ind icators of the extent of financial d evelop m ent. F or exam p le, D e G regorio and G uid otti (1 995) criticize the use of narrow m oney to incom e ratio as a p roxy for financial d evelop m ent. T hey argue that a high level of m onetization (M 1 /G D P ) is the result of financial und erd evelop m ent, while a low level of m onetization is the result of a high d egree of sop histication of financial m arkets, which allow ind ivid uals to econom ize on their m oney hold ings. D e G regorio and G ud otti (1 995) suggest to use a less liq uid m onetary aggregate (M 3 or M 2/G D P ) as a p roxy for financial d evelop m ent. It is exp ected that the b road m oney ratio is p ositively related to growth. M 2Y R atio (M 2Y /Y ): M 2Y d efinition of the m oney includ es the d ep osits in the foreign currency in the national b anking system . A fter financial lib eralization, in a b road er sense, cap ital account lib eralization in 1 989, foreign savers m ay utilize the real return in the countries where the real rate of interest is high. In ad d ition, in an unstab le econom y, the d om estic econom ic unit m ay p refer to hold their assets as foreign currency in the b anking system in ord er to m inim ize the im p act of econom ic shock com ing from the exchange rate risk. It is therefore im p ortant to consid er this ratio as a p roxy for financial d evelop m ent in countries where there is cap ital account lib eralization, which is the case in T urkey p artially after 1 980 and in full after 1 989. 4.3. Construction of Trade Liberalization and Financial D evelopm ent Indexes A s d iscussed in the p revious section, each m easure (p roxy) for trad e lib eralization or financial d evelop m ent cap tures a d ifferent asp ect of the issue and therefore, it is b etter to d evelop a tool to overcom e these p rob lem s. In other word s, the characterization of the relationship am ong alternative p roxies gains im p ortance. In consid ering all these facts, the task is to find out a latent variab le that com b ines d ifferent d im ensions of trad e lib eralization or financial d evelop m ent together and p rovid es a single m easure of trad e lib eralization or financial d evelop m ent. P rincip le com p onent analysis can b e used to com b ine this inform ation in trad e lib eralization and financial d evelop m ent p roxies. T he m ain id ea of p rincip le com p onent analysis is to red uce the d im ensions of a d ata set that consists of a num b er of interrelated variab les, m aking use of the covariance b etween them , while retaining as m uch as p ossib le of the variation p resent in the d ata set (Jolliffe, 1 986). T his is achieved b y the linear transform ation of d ata that are orthogonal to each other. T he m ethod of p rincip le com p onent analysis can b e ap p lied b y using the original values of the d ata or their d eviations from their m eans or stand ard ized variab les. S ince the m ethod is sensitive to the unit of m easurem ent of the d ata, it is b etter to use stand ard ized variab les when the variab les are m easured in d ifferent units. F urtherm ore, consid ering the fact that the p roxies are non-stationary, p rincip le com p onents were estim ated on the d ata m atrix of the d ifference of the logs of the stand ard ized variab les for the p eriod concerned . T he variances of the p rincip le com p onents are the eigenvalues ( ij ) of the variance-covariance m atrix ( ) of the d ata. j T he elem ents of the corresp ond ing eigenvector of the first p rincip le com p onent are the coefficients that will b e used for the linear com b ination of the p roxies. T herefore, the one-d im ensional m easure of trad e lib eralization (or financial d evelop m ent) can b e found as follows: In order to test the joint im pact of trade liberalization and financial developm ent on econom ic growth as discussed theoretically by Blackburn and H ung (1 998), we initially intended to do as Siddiki (2002), who includes two variables for both trade liberalization and financial developm ent in the sam e regression. D ue to the existing high correlation (r = 0,98) am ong trade liberalization and financial developm ent indexes, it m ay not be appropriate to include both at the sam e tim e in a regression. T herefore, we have decided to construct another index that includes both proxies for trade liberalization and financial developm ent, nam ely X /Y , M /Y , O PE N /Y , M 1 /Y , M 1 /M 2, M 2/Y and M 2Y /Y . T his new index (E L ), therefore, involves proxies for both external liberalization and financial developm ent. In a narrow sense, this index (E L ) can be considered as an econom ic liberalization index, which carries instrum ents from both aspects of the issue concerned here. T he E L index is as follows: EL=-0,979L(M1/M2)-0,904L(M1/Y)+0,203L(M2/Y)+0,903L(M2Y/Y) +0,962L(M/Y)+0,97L(X/Y)+0,98L(OPEN) (1 1 ) opt = i =1 T he coefficients of the E L ind ex are consistent with the ab ove find ings. i it (8) 5. Em pirical Results 5.1. D ata set and Time Series Properties of the Variables T he research p eriod is d eterm ined b y the d ata availab ility. T he annual d ata is em p loyed for the T urkish econom y for the p eriod 1 963-2005. T he gross national p rod uct (G N P ) at 1 987 constant p rices is availab le from the web site of the S tate P lanning O rganization. N arrow m oney (M 1 ) and b road m oney (M 2) are taken from the web site of the International F inancial S tatistics (IF S ). M 2Y is collected from the electronic d ata d issem ination system in the C entral B ank in T urkey. T rad e variab les (M and X ) are also taken from the IF S and converted into national currency b y using the exchange rate availab le in the IF S . R ecent d evelop m ents in econom etrics req uires that b efore und ertaking an em p irical analysis, tim e series p rop erties of the d ata in term s of unit root should b e investigated b ecause regression analysis carried out with non-stationary variab les m ay invalid ate m any of the assum p tions of regression analysis. If a tim e series has a unit root, a wid esp read and convenient way to rem ove non-stationarity is b y taking first d ifferences of the relevant variab le. A non-stationary series, which b y d ifferencing d tim es transfers to a stationary one, is called an integrated of ord er d and d enoted as I(d ) (C harem za and D ead m an, 1 997 ). In fact, when a series is integrated of ord er one it m eans that it is not itself stationary, b ut that its first d ifferences are stationary. T he d efinition of the variab les is p resented in T ab le 1 . where opt rep resents the one d im ensional m easure of trad e lib eralization (or financial d evelop m ent) at tim e t, zit is the stand ard ized i th trad e lib eralization (or financial d evelop m ent) p roxy at tim e t, and i is the eigenvector com p onent that corresp ond s to a com p lem entary m easure of i th p roxy. F or trad e lib eralization, three p roxies, nam ely ratio of exp ort to incom e (X /Y ), ratio of im p ort to incom e (M /Y ) and ratio of exp ort p lus im p orts to incom e (O P E N ) are used to ob tain a trad e lib eralization ind ex (T L ): 7 TL=0,9852L(M/Y)+0,9991L(OPEN)+0,98L(X/Y) (9) where L d enotes the logarithm of the following variab les. A s can b e seen from the coefficients of the trad e lib eralization p roxies, they have p ositive im p acts on the trad e lib eralization ind ex. T he ind ex for financial d evelop m ent (F D ) includ es the m onetary aggregates, nam ely M 1 /Y , M 1 /M 2, M 2/Y and M 2Y /Y . T he F D ind ex as follows: FD=-0,905L(M1/Y)+0,23L(M2/Y)+0,941L(M2Y/Y)-0,989L(M1/M2) (1 0) where the all the letters are defined as above. T he coefficient for financial developm ent index indicates that M 1 /Y and M 1 /M 2 are negatively related to the index and the others vice versa. 7 T he results of the construction of the ind exes are availab le up on req uest. N am e of the variab le LPRY LK L SE C TL FD EL D efinition L og of p er cap ita reel incom e L og of gross fixed cap ital form ation as a p roxy for cap ital stock L og of second ary school enrolm ent rate T rad e lib eralization ind ex F inancial d evelop m ent ind ex E conom ic lib eralization ind ex T he results of D F and A D F unit root tests show that the levels of the variab les are not stationary, b ut that their first d ifferences are stationary, with or without the inclusion of a d eterm inistic trend . 5.2. Estim ation of the Long-term Production Function T he next step , in the light of `new' growth theory, is to exam ine the m ultivariate cointegration issue am ong the variab les consid ered . A ccord ingly, a m easure of p hysical cap ital (i.e. gross d om estic cap ital form ation), a m easure of hum an cap ital (i.e. the second ary school enrolm ent rate d efined as the num b er of the stud ents in the second ary school d ivid ed b y total p op ulation) and one of the ind exes, nam ely T L , F D and E L , constructed ab ove will b e includ ed in the em p irical analysis. F ollowing R oub ini and S ala-i M artin (1 992), C oe and M oghad am (1 993), P iazola (1 995), L eigh (1 996), O d ed okun (1 996, 1 999), G hatak, M ilner and U tkulu (1 995, 1 997 ), S id d iki (2002) and L iand and T eng (2006), the augm ented p rod uction function with trad e lib eralization and financial d evelop m ent can b e written as follows: T able 1 . D efinition of the V ariables T he results of the D ickey-F uller (D F ) and A ugm ented D ickey ­ F uller (A D F ) unit root tests for the variab les are p resented in T ab le 2. T he critical values are p rod uced b y the E views 5.0 econom etrics p rogram , which is b ased on the resp onse surfaces in M acK innon (1 991 ). L and D resp ectively d enote the logarithm and d ifference of the variab le in concern. V ariab les LPRY LK L SE C FD TL EL D LPRY D LK D L SE C D FD D TL D EL Intercep t without trend -0.7 50 (0) -1 .090 (0) -1 .97 1 (4) -0.452 (0) -0.293 (0) -0.1 1 0 (0) -7 .035 (0) -5.60 (0) -3.07 (3) -6.230 (0) -5.1 52 (0) -5.22 (0) 5% C ritical V alue -2.93 -2.93 -2.94 -2.93 -2.93 -2.93 -2.93 -2.93 -2.94 -2.93 -2.93 -2.93 Intercep t with trend -2.81 2 (0) -2.31 3 (0) -2.1 23 (0) -3.451 (4) -3.026 (0) -3.494 (4) -6.953 (0) -5.590 (0) -3.1 20 (3) -4.557 (7 ) -5.097 (0) -5.1 85 (0) 5% C ritical V alue -3.52 -3.52 -3.52 -3.53 -3.52 -3.53 -3.52 -3.52 -3.53 -3.54 -3.52 -3.52 R esults Y = f (K , H , X i ) f1 , f 2 , f 3 > 0 (7 ) N ot I(0) N ot I(0) N ot I(0) N ot I(0) N ot I(0) N ot I(0) I(0) I(0) I(0) I(0) I(0) I(0) W here Y is outp ut, K is p hysical cap ital, H is hum an cap ital, X i (i = 1,2,3) d enotes the trad e lib eralization, financial d evelop m ent and the joint im p act of these two com p onents of econom ic lib eralization. T he coefficients for these three variab les are the concerns of this p ap er and it is exp ected that they have p ositive im p acts on econom ic growth. T he rationality of the variab les in the p rod uction function need s som e d etail. T hose variab les exp lained in the p revious section will b e referred to in ord er to avoid rep eating them . S ince it is highly unrealistic to assum e that financial d evelop m ent is the sole or even the m ain d riving force b ehind the growth p rocess, the p otential effects of m ore conventional factors should b e isolated b y includ ing som e variab les. R enelt (1 991 ) argues that it is p ossib le to find a significant relationship b etween growth and m any of the other variab les in the em p irical literature, p articularly in cross-country stud ies. F ollowing R enelt (1 991 ), the p rod uction function includ es sup p ly sid e variab les, nam ely p hysical cap ital, lab our, hum an cap ital, trad e lib eralization and financial d evelop m ent. In ord er to und erstand the true interactions, p otential p olicy variab les are exp licitly m od elled . A ccord ing to S cott (1 992), p hysical cap ital seem s to b e a m uch m ore im p ortant d eterm inant of econom ic growth than neo-classical growth theory suggests. A n increase in investm ent not only raises the rate of econom ic growth, b ut also creates large p ositive external effects (learning effects). P hysical cap ital is ap p roxim ated b y the gross N ot: T he ord er of augm entation in the D ickey-F uller regressions is chosen using the A kaike Inform ation C riterion and the num b ers given in the b rackets in colum ns two and three rep resent the ord er of augm entation. T able 2. D F and A D F T ests for U nit Root fixed cap ital form ation in the em p irical stud ies (F osu, 1 990; G hatak, M ilner and U tkulu, 1 995; P iazolo, 1 995; M ost and van d en B erg, 1 996; A lexand er, 1 997 ; G hura, 1 997 ). T he m ain reason why cap ital form ation is not em p loyed is that there are no annual d ep reciation rates at hand . It is exp ected that the p hysical cap ital should have a p ositive effect on econom ic growth. A sub stantial b od y of recent econom ic theory has em p hasized hum an cap ital as a d eterm inant of econom ic growth. N ot only d oes the new growth theory stress the im p ortance of hum an cap ital (R om er, 1 986; L ucas, 1 988; B arro, 1 991 and 1 998; P iazolo, 1 995; G lom m and R avikum ar, 1 997 ; H wang , 1 998), b ut hum an cap ital augm ented neoclassical m od els as well (M ankiw et. al., 1 992; G ram m y and A ssane, 1 996) . T he ed ucational level of a society serves as a p roxy for the d evelop m ent of hum an cap ital. E d ucation increases the q uality of the lab our force, and therefore, the long-term p rod uction p ossib ility curve shifts outward . A n increase in the ed ucational level has a p ositive effect on econom ic growth. In other word s, a b etter ed ucated lab our force will b e m ore p rod uctive on the job b y req uiring less sup ervision and p ossessing greater initiative in hand ling job -related p rob lem s. T he p rod uction function is estim ated using the Johansen cointegration p roced ure. In p articular, econom ic theory often suggests that the p ath of certain p airs of variab les should not d iverge, at least in the long term , though they m ay d iverge in the short term d ue to seasonal factors. If the variab les continue to d iverge, m arket forces or other instrum ents com m ence to cause them to converge again. In this vein, cointegration m eans that one or m ore linear com b inations of these variab les are stationary even though ind ivid ually they are not. If these variab les are cointegrated , they cannot m ove "too far" away from each other. In other word s, if there is a long-term relationship b etween two or m ore non-stationary variab les, the id ea is that d eviations from this long term p ath are stationary (C harem za and D ead m an, 1 997 ). In contrast, a lack of cointegration suggests that such variab les have no link; they can wand er arb itrarily far away from each other. T he ap p lication of the Johansen cointegration p roced ure (Johansen, 1 988; Johansen and Juselius, 1 992) in the em p irical literature is very com m on. T his m ethod ology em p hasizes the id entification of long-term relationship s, and hence is p articularly ap p rop riate for stud ying the d eterm inants of p otential outp ut. T hese new d evelop m ents in tim e series econom etrics have b een recently exp loited in the growth literature (S erletis, 1 994; Jones, 1 995; P iazolo, 1 995; L eigh, 1 996; A restis and D em etriad es, 1 997 ; C ellini, 1 997 ; H ansson and Jonung, 1 997 ; L au and S in, 1 997 ; B atina, 1 998; H wang , 1 998; R ousseau and W achtel, 1 998; S id d iki, 2002; L iang and T eng, 2006). S tud ies m entioned ab ove utilizing the tim e series p rop erties and cointegration analysis have attem p ted to estab lish a long term relationship b etween the level of the set of the variab les in their em p irical analysis. T he Johansen cointegration p roced ure involves estim ating a V ector A utoregressive M od el (V A R ) such as (H old en and T hom p son, 1 992; C harem za and D ead m an, 1 997 ): Z t = A1 Z t -1 + ... + Ak Z t -k + Dt + ut (1 3) where Z t is a (n x 1 ) vector that contains current and lagged values of n variab les which are each assum ed to b e I(1 ), each A i is an (n x n) m atrix of p aram eters, D t is a vector of I(0) variab les8 and is the vector of rand om errors. H ere, the form ulation of the V A R m od el is of m ajor im p ortance b ecause the results of the cointegration test can b e very sensitive to that form ulation. T here are two m ain interrelated issues that p articularly should b e taken into consid eration. T he first one is to includ e an ap p rop riate lag length to ensure that the resid uals are white noise. T he second is that using too m any lags red uces the p ower of the statistics. T herefore, the choice of the ap p rop riate laglength is im p ortant. T here are several criteria to d eterm ine the ap p rop riate lag-length in the em p irical literature, nam ely the A kaike Inform ation C riteria, the S chwarz and H annan-Q uinn criteria. F ollowing the econom ic d iscussions in the p revious section, the variab les are classified und er three group s. T he first one assesses the im p acts of trad e lib eralization on the p rod uction. T he second one concerns the relationship b etween the financial d evelop m ent and incom e. F inally, the last asserts the joint effect of trad e lib eralization and financial d evelop m ent (econom ic lib eralization) on econom ic growth. T he correlation coefficient b etween trad e lib eralization and financial d evelop m ent ind exes are q uite high (r = 0,98), and therefore two variab les are not includ ed in the sam e regression. F orm ally, the three group s of the variab les are nam ed as m od els and form ed as in T ab le 3: M od el I M od el II M od el III L PRY , L K, L SE C , T L L PRY , L K, L SE C , F D L PRY , L K, L SE C , E L T able 3. D efinition of the V ariables and the Systems E m p irical investigation (carried out E -views econom etric software p rogram m e) starts from an augm ented V A R with four lags on all variab les. T he S chwarz and H annanQ uin criteria showed that, in all m od els, ap p rop riate lag length is eq ual to one (availab le up on req uest). A fter the estab lishm ent and estim ation of the V A R , the cointegration statistics, nam ely m axim um eigenvalue and trace statistics, 8 D t actually rep resents a vector of any variab les that are includ ed in the system to ensure that errors ut are white noise; thus it m ay contain d um m y variab les as well. d evelop ed b y Johansen are ap p lied to test whether there is a the long-term relationship am ong the variab les. T he results are p resented in T ab le 4. H o: rank= r Model I r= = 0 r< = 1 r< = 2 r< = 3 Model II r= = 0 r< = 1 r< = 2 r< = 3 Model III r= = 0 r< = 1 r< = 2 r< = 3 M ax E igen. 35,629* 1 9,21 3 9,242 4,444 38,260* 20,457 1 2,456 3,250 37 ,205* 1 9,61 5 8,628 4,084 5% 28,588 22,299 1 5,892 9,1 64 28,588 22,299 1 5,892 9,1 64 28,588 22,299 1 5,892 9,1 64 T race 68,529* 32,899 1 3,686 4,444 7 4,425* 36,1 64* 1 5,7 07 3,250 69,534* 32,328 1 2,7 1 2 4,084 5% 54,07 9 35,1 92 20,261 9,1 64 54,07 9 35,1 92 20,261 9,1 64 54,07 9 35,1 92 20,261 9,1 64 (E q u atio n 1 4 ). In ad d itio n , fin an cial d ev elo p m en t h as p o sitiv ely co n trib u ted to eco n o m ic g ro w th (eq u atio n 1 5 ). F in ally, alth o u g h th e m ag n itu d e o f its co efficien t is v ery sm all, th e v ariab le fo r eco n o m ic lib eraliz atio n (E L ), w h ich cap tu res th e asp ects o f b o th trad e lib eraliz atio n an d fin an cial d ev elo p m en t, also h as a p o sitiv e effect o n th e g ro w th rate. H o w ev er, p h ysical cap ital in M o d els II an d III (L K in E q u atio n 1 5 an d 1 6 ) seem s to b e in sig n ifican t. T h e em p irical an alysis sh o w s th at fin an cial d ev elo p m en t (F D ) an d trad e lib eraliz atio n (T L ) sep arately h av e p o sitiv e co n trib u tio n s to eco n o m ic g ro w th in T u rkey. In ad d itio n , th e jo in t im p act o f trad e lib eraliz atio n an d fin an cial d ev elo p m en t in term s o f eco n o m ic lib eraliz atio n (E L ) also p o sitiv ely in flu en ce eco n o m ic g ro w th . A s arg u ed b y th e en d o g en o u s g ro w th literatu re, h u m an cap ital also p o sitiv ely affects in co m e. 6. Con clusion W ith th e em erg en ce o f th e n ew g ro w th th eo ries, th e im p act o f eco n o m ic p o licy o n eco n o m ic g ro w th h as b een a h o tly d eb ated issu e. In o th er w o rd s, w h eth er eco n o m ic p o lices in a co u n try in flu en ce th e eco n o m ic g ro w th is an em p irical issu e. T h is p ap er em p irically in v estig ated w h eth er trad e lib eraliz atio n an d fin an cial lib eraliz atio n h av e h ad an y sig n ifican t im p act o n eco n o m ic g ro w th in T u rkey w h ich , as a d ev elo p in g eco n o m y, h as w itn essed an u n p reced en ted stag ed refo rm attem p t in v o lv in g extern al (trad e) an d in tern al (fin an cial) lib eraliz atio n , esp ecially after th e 2 4 J an u ary D ecisio n s fo llo w in g th e eco n o m ic crisis in 1 9 8 0 . In T u rkey, eco n o m ic lib eraliz atio n in term s o f trad e an d th e fin an cial secto r w as at th e h eart o f th e stab iliz atio n p ro g ram m e em p lo yed in 1 9 8 0 an d h as co n stitu ted an in teg ral p art o f th e eco n o m ic p o licies sin ce th en . T h e th eo retical co re o f th e eco n o m ic refo rm p ro g ram m e is to p ro v id e efficien cy in th e allo catio n o f scarce reso u rces an d to m o b iliz e u n p ro d u ctiv e reso u rces in to in v estm en t an d h en ce p ro m o te eco n o m ic g ro w th . T o test th e im p act o f trad e lib eraliz atio n an d fin an cial d ev elo p m en t o n T u rkish eco n o m ic g ro w th , th ree altern ativ e m easu res (in d exes) w ere d ev elo p ed b y m akin g u se o f th e p rin cip al co m p o n en ts an alysis, n am ely trad e lib eraliz atio n , fin an cial d ev elo p m en t an d , in a sen se, eco n o m ic lib eraliz atio n p ro xies. T h e em p irical resu lts, o b tain ed b y em p lo yin g th e m eth o d s o f th e tim e series eco n o m etrics fo r th e p erio d 1 9 6 3 2 0 0 5 , sh o w ed th at trad e lib eraliz atio n an d fin an cial d ev elo p m en t p o sitiv ely co n trib u tes to eco n o m ic g ro w th . F u rth erm o re, th e jo in t im p act o f trad e lib eraliz atio n an d fin an cial d ev elo p m en t in term s o f eco n o m ic lib eraliz atio n o n eco n o m ic g ro w th is also sig n ifican t in T u rkey. T able 4. Johansen Cointegration T est Results B oth trace and m axim um eigenvalue statistics showed that there is at least one cointegrating vector for each m od el. T race statistics ind icated that there are two cointegrating vectors in M od el II. F or the sam e m od el, the m axim um eigenvalue statistic resulted in one cointegrating vector. T herefore, one cointegrating vector for the M od el II is assum ed in the further analysis. T he corresp ond ing cointegrating vectors for each m od el resp ectively p resented as follows (stand ard errors are given in the p arenthesis): (1 4) (1 5) (1 6) A s far as th e em p irical resu lts are co n cern ed , seco n d ary sch o o l en ro lm en t rate, fin an cial d ev elo p m en t an d trad e lib eraliz atio n are statistically sig n ifican t. T h e sig n o f th e v ariab les co n cern ed (T L , F D an d E L ) is as arg u ed in th e th eo retical d iscu ssio n ab o v e. P articu larly, h u m an cap ital accu m u latio n (L S E C ) in th ree m o d els (E q u atio n 1 4 , 1 5 , 1 6 ) is sig n ifican t an d in flu en ces eco n o m ic g ro w th p o sitiv ely. T h e im p act o f trad e lib eraliz atio n (T L ) o n p er cap ita in co m e is p o sitiv e http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png South East European Journal of Economics and Business de Gruyter

Trade Liberalization, Financial Development and Economic Growth in The Long Term: The Case of Turkey

Loading next page...
 
/lp/de-gruyter/trade-liberalization-financial-development-and-economic-growth-in-the-Q4VQIHlZmZ
Publisher
de Gruyter
Copyright
Copyright © 2008 by the
ISSN
1840-118X
DOI
10.2478/v10033-008-0012-x
Publisher site
See Article on Publisher Site

Abstract

T he determinants of economic growth have been a much debated theoretical issue in the literature, especially after the endogenous growth theory of the late 1 980s. T his new theory highlights the importance of economic policies that lead to an increasing rate of return. In particular, it is argued that human capital, trade liberalization and financial development may play very important roles in the determination of economic growth. T his paper tries to empirically estimate the joint impacts of trade liberalization and financial development on economic growth for the period 1 9602 004. Instead of using common proxies for the issue, principal components analysis is employed to develop better measures (indexes) for trade liberalization, financial development and the joint effects of both. T he empirical results obtained from the J ohansen co-integration procedure show that trade liberalization, financial development and the joint impacts of both positively contributed to economic growth in T urkey for the period 1 963-2 005. Keywords: Trade liberalization, Financial development, Economic growth, Co-integration, Endogenous growth theory, Turkey. J EL: C32 , E44, F 1 3, O 1 1 , O 1 6. DOI: 10.2478/v10033-008-0012-x 1. Introduction The im pact of financial and trade regim es on econom ic growth have been a hotly debated theoretical issue, particularly after the em ergence of the endogenous (new) growth theory during the last two decades. Contrary to the neo-classical growth theory, the new growth theory im plies that the econom ic policies followed in a country m ay have a significant im pact on the long-term rate of econom ic growth. To this end, m ost developing countries that form erly followed restrictive econom ic policies have started liberalizing their trade and financial sectors in order to increase econom ic growth in the 1980s. The m ain argum ent for this policy change was that both trade and financial liberalization policies reduce inefficiency in the production process and positively influence econom ic growth. Following this line of reasoning, Turkey, as a developing econom y, has witnessed an unprecedented and staged reform attem pt involving external (trade) and internal (financial) liberalization, especially after the 24 January Decisions following the econom ic crisis in 1980. In Turkey, econom ic liberalization in term s of trade and financial sector was at the heart of the stabilization program m e Novem ber 2008 *Kar Faculty of Economics and Administrative Sciences, Kahramanmara Sütçü mam University, Kahramanmara/TURKEY e-mail: mkar@ksu.edu.tr *Peker Faculty of Nazilli Economics and Administrative Sciences Adnan Menderes University, Aydin/TURKEY e-mail: ottopeker@yahoo.com *Kaplan Faculty of Economics and Administrative Sciences, Nide University, Nide/TURKEY e-mail: mkaplan@nigde.edu.tr em ployed in 1980 and constituted an integral part of the econom ic policies since then. The theoretical core of the econom ic reform program m e is to provide efficiency in the allocation of scarce resources. In the em pirical literature, the im pact of trade liberalization 1 and financial developm ent2 on econom ic growth has been separately exam ined for Turkey. However, there has been theoretical literature that highlights the joint im pact of these issues on econom ic growth. The joint im pact of both variables is initially highlighted in the Roubini and Sala-i Martin (1991) m odel, which has extended the Barro (1991) growth m odel by incorporating both factors. The inclusion of both trade and financial variables in the Barro (1991) m odel showed that financial and trade variables m ay play a very im portant role in econom ic growth. Recently, Blackburn and Hung (1998) have presented a theoretical analysis of the joint im pact of both financial developm ent and trade liberalisation on econom ic growth. Using endogenous growth theory, their m odel predicts that econom ic growth rates in the presence of financial interm ediation tend to be higher than those under direct lending and borrowing. The m odel also predicts that both financial developm ent and trade liberalization jointly facilitate the rate of econom ic growth by decreasing redundant research efforts and increasing m arkets for new products. Therefore, this paper em pirically exam ines the im pacts of trade liberalization and financial liberalization and/ or developm ent on econom ic growth in Turkey by using a sim ple endogenous growth fram ework and by m aking use of new developm ents in tim e series techniques for the period 1963-2005. This paper is distinguished from the earlier existing literature in two aspects. First of all, this paper tries to assess the joint im pact of trade liberalization and financial developm ent on econom ic growth. Second, there are different m easures for trade liberalization and financial developm ent in the literature and the existing studies em ploy only one of these proxies in their analysis. However, the trade and financial liberalization affect econom ic growth through different channels and each proxy captures a single aspect of the issue. In order to overcom e this problem , three com posite indexes, for trade liberalization, for financial developm ent 1 The relationship between growth and trade liberalization is usually exam ined in the em pirical literature in two different lines for Turkey: The first line of the existing em pirical research tries to assess the im pact of openness on econom ic growth (Conway, 1987; Greenaway and Sapsford, 1994, 1995; Ghatak, Milner and Utkulu, 1995; Subasat, 2002; Utkulu and Ozdem ir, 2005, Hilm i and Safa, 2007). The second line of the em pirical works exam ines the relationship between trade liberalization and the total factor productivity (Krueger and Tuncer, 1980, 1982; Nishim izu and Robinson, 1986; Foroutan, 1991; Ozm ucur and Karatas, 1994; Filiztekin 2000). 2 With the em ergence of financial liberalization hypothesis, the em pirical studies tried to assess the im pact of financial liberalization on econom ic growth by sim ple regressing growth rate on the real interest rate (Fry, 1978, 1979, 1980). Furtherm ore, the relationship between financial developm ent and econom ic growth is analyzed in term s of causality issue in a bivariate m odel (Kar and Pentecost, 2000; Unalm is, 2002; Ozatay and Sak, 2002). In addition, Yeldan (1997), Lewis (1992) and Karabulut and Dem iroz (2002) exam ined the im pact of financial liberalization/developm ent on econom ic growth in the fram ework of the CGE m odels. Finally, Guncavdi and Kucukciftci (2005) investigate the role of financial reform s on econom ic growth with a m ethodology based on the Leontief's input-output m odel and conclude that the production sector of the econom y has increasingly becom e independent from the use of financial services produced by the banking system in the post-reform period. and for a narrow sense econom ic liberalization, are constructed by applying principal com ponents analysis. The structure of this paper is as follows: the next section presents the theoretical linkages am ong trade liberalization, financial developm ent and econom ic, growth and introduces the aggregate production function to be estim ated. Section three gives brief inform ation about the Turkish econom y, particularly highlighting the developm ents in trade and financial sector. In section four, the m easures for trade liberalization and financial developm ent are reviewed and three indexes are constructed by em ploying principal com ponents analysis. In section five, the em pirical results are presented using econom etric techniques and the outcom e of the longterm production function is interpreted in detail. The paper finishes with a conclusion. 2.Theoretical Framework: Trade Liberalization, Financial Development and Economic Growth With the em ergence of the endogenous growth theories in 1980s, the relationship between econom ic policy and growth becam e a highly debated issue. In the theoretical literature, discussions are focused on different channels through which econom ic policy affects econom ic growth. In this section, different m odels of growth will be discussed to provide a fram ework of thought that helps to understand the im pacts of each link between policy and growth. In other words, each channel through which econom ic policy affects growth has different im plications for growth in different m odels. There are two com peting theoretical fram eworks in the growth literature, nam ely neo-classical and endogenous growth theories. The m ain differences between them are whether the policy change has a longterm effect on the growth rate. On the one hand, the neoclassical theory argues that a policy change has no effect on long-term econom ic growth and, on the other hand, that endogenous growth theory shows a policy change in econom y does m atter. The endogenous growth theory is a reaction to the traditional Neo-Classical growth m odels3 , represented by, am ong others, Solow (1956). This new approach to growth theory has sought to supply the m issing explanation of longterm growth. In essence, this approach provides a theory of technical progress, one of the central m issing elem ents of the neo-classical m odel. In other words, endogenous growth theories seek to discover what lies behind the exogenous rate of technical progress and hence a country's growth rate. Endogenous growth theory recognizes that technological change occurs as a result of the efforts of profit-m axim izing firm s to invent new blueprints, and that technological progress is an endogenous outcom e of econom ic activity. 3 For a recent em pirical and theoretical review of the literature on growth, see Renelt (1991), Herm es (1994), Levine (1997, 2001), Thirlwall (2000), Favara (2003), Auerbach and Siddiki (2004), Subasat (2002), Winters (2004). SEE Journal The crucial distinction between `old' and `new' growth theories is that the form er utilizes the assum ption that returns to the capital stock is dim inishing, while the latter argues that returns to capital itself or, in a wider sense, to the stock of physical and hum an capital form ation is constant or increasing (Sala-i Martin, 1990a). This then im plies that those variables that lead to non-decreasing returns drive the growth rate. Num erous candidates have been recom m ended as the source of non-decreasing returns: particularly, the stock of hum an capital Lucas (1988); accum ulated capital, Rebelo (1991); research and developm ent, Rom er (1986, 1990); or public infrastructure investm ent (Barro, 1991). Thus, endogenous growth m odels highlight sectors of the econom y that influence the growth path of an econom y. This can be sim ply shown in a Robelo-type production function, known as the AK m odel. Most of the endogenous growth m odels can be viewed as extensions or m icro-foundations of the AK m odel (Sala-iMartin, 1990b). Rebelo (1991) form ulated the sim ple form of the endogenous growth m odel, which has since been widely used in em pirical analysis. The AK m odel takes its nam e from its production function. In its original form , the m odel setting involves dynam ic m axim ization. In this section, we will m ake the further assum ption of a constant savings rate. This assum ption, however, does not change the m ain conclusions and intuitions of the m odel. In the AK m odel, the production function takes the following form : the cap ital accum ulates. T his can b e shown easily: after sub stituting the value of investm ent into eq uation (3) and then d ivid ing b oth sid es b y K t -1 and taking the logarithm of b oth sid es, the resulting eq uation will b e: K log t = log[sA +(1 - )] K t -1 F or sm all values of s, A and (4) can written as: and (4) sA > , eq uation log K t = sA - (5) T his eq uation says that the rate of growth of cap ital stock is constant if tariff rates are constant. A fter taking the logarithm and d erivative of the p rod uction function and sub stituting the value of the eq uation of m otion of the cap ital from eq uation (5) and the value of return to cap ital from the eq uation (2), the longterm rate of growth of outp ut can b e written as follows: log Yt = log K t = s -s (6) Yt =A K t (1) where Yt represents output, K t is capital stock at tim e t and A is som e positive constant. This form ulation of the production function m eans that there are constant returns to capital accum ulation. It is also im portant to note that A is equal to the return to investm ent in this m odel. As will be explained in the next section, trade policy prim arily affects the rate of return of capital and hence growth. Therefore, A can be written as a function of trade policy () as, A= (2) Equation (2) indicates that the rate of return of capital is a negative function of trade policy. The accum ulation of capital is form ulated as: K t = I t -1 + (1 - ) K t -1 and I t = sYt (3) where s is the investm ent rate and is the depreciation rate. Both are assum ed constants, and investm ent at tim e t ( I t ) is equal to the savings in the econom y. The special form ulation of the production function in the AK m odel (equation 1) im plies that the m arginal product of each unit of capital is always equal to A. It does not decline as Novem ber 2008 F rom eq uation (6), it is ob vious that the rate of growth of the econom y is d ecreasing with tariff rates and increasing with saving rates. H ence, any econom ic p olicy that increases the return to investm ent will p erm anently increase the rate of growth of the econom y. A lm ost allend ogenous growth literature has concentrated on the d eterm inants of the return to investm ent, A , and how p olicy change affects it (S ala-i-M artin, 1 990a). E q uation (6) has two m ain im p lications in term s of econom ic p olicy change. E conom ic p olicies in the trad e and financial sectors will have a long-term effect on econom ic growth. W hile trad e p olicy affects the econom ic growth through the change on tariff rates, the saving rates are influenced b y a well-functioning financial system . T he theoretical linkages am ong trad e lib eralization and financial lib eralization and econom ic growth can b e exp lained as follows. T rad e p olicy in term s of tariff red uction or elim ination of restrictions on trad e m ight have im p acts on the growth through several channels. If op enness is to affect econom ic p erform ance, it m ust have an interm ed iate effect on one or all of the following: (i) allocation of factors of p rod uction across sectors - the allocation effect (Y oung, 1 991 ; R ed d ing, 1 997 ; G rossm an and H elp m an, 1 992); (ii) op enness will increase com p etition in the d om estic econom y and hence p rod uctivity - the im p ort d iscip line hyp othesis (G reenaway and M ilner, 1 993; A ghion, D ewatrip ont and R ey, 1 997 ; A ghion, H arris and V ickers, 1 997 ; A ghion and H owitt, 1 996); (iii) op enness enlarges the m arket for d om estic p rod ucers, which they can take ad vantage of - the scale econom ies (T aylor, 1 994; G rossm an and H elp m an, 1 991 ); (iv) op enness increases the num b er of inp uts that have no d om estic sub stitutes and thus lead s to a higher cap acity for utilization and p rod uctivity - the availab ility of inp uts (N ishim izu and R ob inson, 1 986; Q uah and R auch, 1 990; R ivera-B atiz, and R om er, 1 991 ; G rossm an and H elp m an, 1 992); and , finally, (v) the flow of knowled ge across sectors and countries the sp illover effect (F ed er, 1 982; G rossm an and H elp m an, 1 992). O n the other hand , the role of financial sector in econom ic d evelop m ent has long b een one of the hotly d eb ated issues am ong econom ists (S chum p eter, 1 91 1 ; G old sm ith, 1 969; P atrick, 1 966; M cK innon, 1 97 3; S haw, 1 97 3). W ith the em ergence of the end ogenous growth theory, several stud ies have attem p ted to show how the op eration and p olicies of the financial sector m ay affect the rate of econom ic growth (G reenwood and Jovanovic, 1 990; B encivenga and S m ith, 1 991 ; K ing and L evine, 1 993a, 1 993b ; R oub ini and S ala-i M artin, 1 992; P agano, 1 993; L eigh, 1 996; D em etriad es and H ussein, 1 996; A restis, 2005; S id d iki, 2002; A uerb ach and S id d iki, 2004; A restis, 2005; L iang and T eng, 2006). F inance can influence growth in an end ogenous growth m od el through increasing the savings rate (B encivenga and S m ith, 1 991 ), b y increasing the returns on investm ent (G reenwood and Jovanovic, 1 990), and b y increasing hum an cap ital accum ulation. F rom a m acroeconom ic or aggregate p rod uction function p oint of view, all this m eans that econom ies that are d evelop ed m ore financially will b e ab le to transform a given am ount of inp uts, K , into larger am ount of outp ut, Y . T his is why the p rod uction function is an increasing function of the financial d evelop m ent of the econom y (R oub ini and S ala-I M artin, 1 992). term econom ic d evelop m ent p lan. T his p rop osal was for three five-year-p lans, the first of which b egan in 1 963. T he econom ic d evelop m ent p lans were aim ed at p rod ucing a well-b alanced econom y with p rogress in agriculture and ind ustry, thereb y m aking the m ost of the availab le resources in the T urkish econom y. T he m ain features of this p eriod are that the econom ic p olicies carried out within the d evelop m ent p lans were characterized as interventionist and p rotectionist. A ccord ingly, p olicies were m ainly d esigned to p rotect d om estic ind ustry from foreign com p etition and to increase the governm ent control over the allocation of resources and p rod uction of good s. T hese econom ic p olicies im p lem ented in the early 1 960s were p ursued for ab out twenty years, until 1 980, when the T urkish econom y found itself in a m ajor crisis. In 1 980, a new stab ilization p olicy was accep ted and the etatism and im p ort sub stitution p olicies were switched to an exp ort-oriented ind ustrialization p olicy b ased on a m arket m echanism . T he p olicy p ackage p ut into effect in 1 980 and reinforced in the following years was m ore than just a stab ilization and ad justm ent p ackage; it also m arked a shift in d evelop m ent strategy from inward orientation to outward orientation (Y ild izoglu and M argulies, 1 988; S enses, 1 984; O nis, 1 986; D ervis and P etri, 1 987 ). T rad e and financial lib eralization were the m ain p olicy tools in this stab ilization p rogram m e. 3.2. Trade Liberalization in Turkey A fter p ursuing the "im p ort sub stituting ind ustrialisation strategy (IS I)" as a d om inant ind ustrialization strategy in the 1 950s, 1 960s, and 1 97 0s, T urkey switched to an outward -oriented ind ustrialization strategy with the IM F sup p orted stab ilization p rogram m e that was introd uced to resum e growth following the econom ic crisis in 1 980. R ap id exp ort growth was one of the m ain ob jectives of the 1 980 stab ilization p rogram m e to im p rove the huge trad e d eficit, restore international cred itworthiness and estab lish the cred ib ility of lib eralization reform s at hom e. A variety of incentives were introd uced to p rom ote m anufactured exp orts. T hese incentives includ ed tax reb ates, cred it sub sid ies, and foreign exchange allocated for the im p ort of interm ed iate p rod ucts. T he success of the lib eralization p rocess in the 1 980s p rom p ted the governm ent to p ursue further lib eralization in the 1 990s. T herefore, T urkey lib eralized her im p ort regim e b y ab olishing the d ep osit req uirem ent for im p orts and the im p ort licensing system in the early 1 990s. A ccom p anying agreem ents with the W orld T rad e O rganisation (W T O ) in 1 994 have significantly contrib uted to the lib eralization of the im p ort regim e. A s a m em b er of the W T O , T urkey has ad op ted the rules and p roced ures governing the m ultilateral trad ing system and entered into negotiations with several E astern and C entral E urop ean, M ed iterranean and B altic countries to conclud e free trad e agreem ents. 3. A Brief Review of Turkish Econom y 3.1. Macroeconom ic D evelopm ents It is very com m on to exam ine m acroeconom ic d evelop m ent in the T urkish econom y und er three sub -p eriod s, 1 923-1 960, 1 960-1 980, 1 980 and after, d istinguished not only b y d ifferent structural cond itions, b ut also b y the governm ent's resp onse to those cond itions. T here was heavy state intervention form ulated as "etatism " b efore 1 960 (O kyar, 1 965). E tatism , an econom ic p olicy excessively controlled b y the S tate, b ecam e unp op ular and heavily criticized am ong T urkish acad em ics and b usinessm en. W ith the changes in the p olitical structure in term s of the m ulti-p arty system , there was a p olicy change toward s lib eralization in the p eriod 1 950-53. H owever, with the d eterioration of m acroeconom ic structure, the etatist econom ic p olicies were re-initiated after attem p ts at p olicy change. A fter the 1 960 m ilitary coup , T urkey entered into a p eriod of d evelop m ent p lans. In S ep tem b er 1 960, the S tate P lanning O rganisation (S P O ) was estab lished to stud y the T urkish econom y and to p rop ose and enforce a long28 In 1 996, T urkey entered into a new era b y signing a C ustom s U nion A greem ent with the E urop ean U nion (E U ). A fter the lib eralization p rogram m e in 1 980, this was the second m ost im p ortant d evelop m ent affecting the T urkish econom y as a whole. E xcep t for sensitive p rod ucts, m ainly m otor vehicles, footwear, and furniture, T urkey lifted all tariff and non-tariff b arriers for m anufacturing p rod ucts originating from the E U . T urkey also ad op ted the E U 's C om m on E xternal T ariff for good s im p orted into T urkey from third countries. T his req uired a further lib eralization of her tariff regim e, since T urkish p rotection rates were higher overall than the C om m on E xternal T ariff (H arrison et al., 1 996; T ogan, 1 997 ). M oreover, T urkey had to red uce her tariffs to countries that signed a P referential T rad e A greem ent (P T A ) with the E U . 3.3. Financial Liberalization in Turkey P rior to 1 980, T urkey was a typ ical exam p le of highly restricted and segm ented financial m arkets (A kyuz, 1 990). Interest rates were d eterm ined institutionally and kep t at artificially low levels. B y the end of the 1 97 0s, real interest rates b ecam e highly negative d ue to the acceleration of inflation (F ry, 1 97 9). T he state owned b anks were d om inant institutions in the T urkish financial system (F ry, 1 97 9). E ntry into the b anking sector (d om estic and foreign) was restricted . S ince then, T urkey has b een exp eriencing a lib eral ap p roach to its financial m arkets as a key com p onent of the newly ad op ted growth-oriented structural ad justm ent p rogram since 1 980 (A ricanli and R od rik, 1 990). W hen the authorities lifted the ceilings on p ersonal tim e d ep osit rates and lend ing rates were ab olished 4 . A t the tim e it was consid ered a "m ajor step in d eregulation of interest rates which b reaks a p ractice that has b een in force som e 50 years" (W olff, 1 987 :1 04). T his p olicy change was not very sm ooth and the reluctant b ehaviour of the financial institutions req uired the authorities to intervene into the sector two years after the im p lem entation of financial lib eralization and to d eterm ine interest rate for a p eriod . 5 A nother d evelop m ent, in the first half of the 1 980s, was that resid ents (and non-resid ents) were allowed to op en foreign exchange d ep osits in com m ercial b anks (R ittenb erg, 1 988). T he essential regulation was finally initiated in 1 985, when the new B anking L aw was enacted . T he law introd uced new regulations in term s of p rovision for a m inim um cap ital and a cap ital ad eq uacy ratio. T he ownership structure of b anks was also regulated . F urtherm ore, a B ank S up ervision unit at the C entral B ank b ecam e op erational in 1 986. T hus, five years after the initial lib eralization of d om estic interest rates, an ad eq uate regulatory and institutional fram ework 4 T he d evelop m ents in financial sector since 1 980 are exp lained in m ore d etail A kyuz (1 990), C osan and E rsel (1 986), Inselb ag and G ultekin (1 988), C izre-S akallioglu and Y eld an (2000), O zatay and S ak (2002) and R ittenb erg (1 988). 5 A tiyas (1 990) exp lains in d etail of the resp onse of p rivate sector to financial lib eralisation. was d efined and b ecam e op erational. A fter the lib eralization of the cap ital account in 1 989 and estab lishm ent of the sup ervisory and regulatory unit at the C entral B ank, it was b elieved that an ad eq uate regulatory and institutional fram ework was finally d efined and b ecam e functional. H owever, the crises of N ovem b er 2000 and F eb ruary 2001 have shown that these d evelop m ents were not sufficient to have stab le, efficient and wellestab lished financial m arkets in T urkey. T hese crises led not only to the estab lishm ent of a new institution, nam ely the S up ervisory and R egulatory B oard of B anking , which aim ed at restructuring the financial system in 2000, b ut also to the d evelop m ent of new p olicies that p rop osed an efficient and effective financial m arket. It can b e argued that the T urkish financial system finally had an ind ep end ent sup ervisory and regulatory b od y, 20 years after b eginning to im p lem ent lib eralization p olicies in the financial sector. 4. Measurem ent of Trade Liberalization and Financial D evelopm ent 4.1. Measurem ent of Trade Liberalization R esearchers in the recent em p irical literature concentrate on find ing reliab le p roxies of trad e lib eralization. H owever, the share of exp ort as a p ercentage of incom e, the share of the im p ort as a p ercentage of incom e, the share of exp ort p lus im p orts (trad e volum e) as a p ercentage of the incom e and tariffs constitute very com m on p roxies for trad e lib eralization in the em p irical literature. In this article, the following p roxies of trad e lib eralization are em p loyed in the em p irical analysis. E xp ort to G D P ratio (X /G D P ): T he first theoretical channel that links op enness to econom ic p erform ance goes through the allocation of resources. A ccord ing to this argum ent, op ening up to international trad e b rings ab out reallocation of resources accord ing to com p arative ad vantages (G rossm an and H elp m an, 1 992, Y oung, 1 991 ). S ince the d irect effect of the allocation of resources is ob served on the level of exp orts, the share of exp orts in total p rod uction can b e used to rep resent this d im ension of op enness. In ad d ition, the share of exp orts in p rod uction can b e used as a p roxy of op enness to cap ture the d im ension of op enness related to scale econom ies and the availab ility of inp uts. Im p ort to G D P ratio (M /G D P ): T he im p ort share in total p rod uction can b e used as an op enness p roxy characterizing the d im ension of op enness related to increased international com p etition. It also rep resents the allocation effect of op enness since the im p orts of those sectors that have com p arative d isad vantages will increase following trad e lib eralization. F oreign trad e to G D P ratio (X + M /G D P ): T he share of the total of exp orts and im p orts in total p rod uction p rovid e the p roxy that rep resents the technology sp illover d im ension of op enness. O p enness to trad e facilitates access to the technological inform ation in the world (G rossm an and H elp m an, 1 992, C hp -9), which creates technological sp illover through im p orts as well as exp orts. 4.2. Measurem ent of Financial D evelopm ent O ne of the m ost d ifficult asp ects of em p irically investigating the relationship b etween financial d evelop m ent and econom ic growth is the m easurem ent of "financial d evelop m ent". H owever, the p ractitioners are forced to form a well-d efined set of m easures of financial d evelop m ent b y the availab ility of d ata at hand . T he p roxies p rop osed for m easuring the level of financial d evelop m ent are b asically chosen from the m onetary and cred it aggregates in an econom y. T he rationale for the inclusion of a wid e range of p roxies is to m axim ize the inform ation on financial d evelop m ent. In other word s, d iverse aggregates should b e ab le to catch d ifferent functions of the financial m arkets. In this article, the following p roxies6 for financial d evelop m ent are em p loyed in the em p irical analysis. N arrow M oney R atio (M 1 /Y ): In the ab sence of the financial sector econom ic agents have to hold their financial assets at hand . T his m eans that the fund s for investm ent will b e kep t out of the financial sector in the econom y. W ith financial d evelop m ent, the ratio of narrow m oney to incom e will d ecrease. N arrow M oney B road M oney R atio (M 1 /M 2): In a fragm ented financial sector, econom ic units m ay p refer to hold their fund s out of the financial sector to rem ain liq uid . H owever, d evelop m ents in the financial system in term s of not only organizations such as b anks b ut also instrum ents in this sector m ay lead p eop le to p ut their m oney in the b anking sector, through which investm ent can b e carried out. In short, with financial d evelop m ent, d ep osits in the b anking sector m ay b e increased and as a result the M 1 / M 2 ratio will d ecrease. B road M oney R atio (M 2/Y ): M onetary aggregates also p rovid e an alternative set of variab les to m easure the extent of financial d evelop m ent (D e G regorio and G uid otti, 1 995; G aletovic, 1 996; L ynch, 1 996). In the literature, the com m only used m easure of financial d evelop m ent is a ratio of som e b road m easure of the m oney stock, usually M 2, to the level of nom inal incom e (G elb , 1 989; K ing and L evine, 1 993a, 1 993b ; M urind e and E ng , 1 994a, and 1 994; L yons and M urind e, 1 994; D em etriad es and H ussein, 1 996; A restis and D em etriad es, 1 997 ; K wan et. al., 1 998). T his sim p le ind icator m easures the d egree of m onetization in the econom y. T he m onetization variab le is d esigned to show the real size of the financial sector of a growing econom y. M oney p rovid es valuab le p aym ent and saving services. T he `narrow m oney' stock b est reflects the form er 6 F or a m ore com p rehensive set of m easures of financial d evelop m ent, see L ynch (1 996). and `b road m oney' the latter. N arrow m oney b alances should rise in line with econom ic transactions, b ut b road m oney should rise at a faster p ace, if financial d eep ening is occurring (L ynch, 1 996). It is argued that the use of m onetary aggregates as a p roxy for the d egree of financial d evelop m ent m ight also p resents p rob lem s (D e G regorio and G uid otti, 1 995; K ing and L evine, 1 993a). K ing and L evine (1 993a) note that d ifferent d efinitions of m onetary aggregates m ay act as p roxies for d ifferent roles of financial interm ed iation. In som e cases m onetary aggregates m ay b e very p oor ind icators of the extent of financial d evelop m ent. F or exam p le, D e G regorio and G uid otti (1 995) criticize the use of narrow m oney to incom e ratio as a p roxy for financial d evelop m ent. T hey argue that a high level of m onetization (M 1 /G D P ) is the result of financial und erd evelop m ent, while a low level of m onetization is the result of a high d egree of sop histication of financial m arkets, which allow ind ivid uals to econom ize on their m oney hold ings. D e G regorio and G ud otti (1 995) suggest to use a less liq uid m onetary aggregate (M 3 or M 2/G D P ) as a p roxy for financial d evelop m ent. It is exp ected that the b road m oney ratio is p ositively related to growth. M 2Y R atio (M 2Y /Y ): M 2Y d efinition of the m oney includ es the d ep osits in the foreign currency in the national b anking system . A fter financial lib eralization, in a b road er sense, cap ital account lib eralization in 1 989, foreign savers m ay utilize the real return in the countries where the real rate of interest is high. In ad d ition, in an unstab le econom y, the d om estic econom ic unit m ay p refer to hold their assets as foreign currency in the b anking system in ord er to m inim ize the im p act of econom ic shock com ing from the exchange rate risk. It is therefore im p ortant to consid er this ratio as a p roxy for financial d evelop m ent in countries where there is cap ital account lib eralization, which is the case in T urkey p artially after 1 980 and in full after 1 989. 4.3. Construction of Trade Liberalization and Financial D evelopm ent Indexes A s d iscussed in the p revious section, each m easure (p roxy) for trad e lib eralization or financial d evelop m ent cap tures a d ifferent asp ect of the issue and therefore, it is b etter to d evelop a tool to overcom e these p rob lem s. In other word s, the characterization of the relationship am ong alternative p roxies gains im p ortance. In consid ering all these facts, the task is to find out a latent variab le that com b ines d ifferent d im ensions of trad e lib eralization or financial d evelop m ent together and p rovid es a single m easure of trad e lib eralization or financial d evelop m ent. P rincip le com p onent analysis can b e used to com b ine this inform ation in trad e lib eralization and financial d evelop m ent p roxies. T he m ain id ea of p rincip le com p onent analysis is to red uce the d im ensions of a d ata set that consists of a num b er of interrelated variab les, m aking use of the covariance b etween them , while retaining as m uch as p ossib le of the variation p resent in the d ata set (Jolliffe, 1 986). T his is achieved b y the linear transform ation of d ata that are orthogonal to each other. T he m ethod of p rincip le com p onent analysis can b e ap p lied b y using the original values of the d ata or their d eviations from their m eans or stand ard ized variab les. S ince the m ethod is sensitive to the unit of m easurem ent of the d ata, it is b etter to use stand ard ized variab les when the variab les are m easured in d ifferent units. F urtherm ore, consid ering the fact that the p roxies are non-stationary, p rincip le com p onents were estim ated on the d ata m atrix of the d ifference of the logs of the stand ard ized variab les for the p eriod concerned . T he variances of the p rincip le com p onents are the eigenvalues ( ij ) of the variance-covariance m atrix ( ) of the d ata. j T he elem ents of the corresp ond ing eigenvector of the first p rincip le com p onent are the coefficients that will b e used for the linear com b ination of the p roxies. T herefore, the one-d im ensional m easure of trad e lib eralization (or financial d evelop m ent) can b e found as follows: In order to test the joint im pact of trade liberalization and financial developm ent on econom ic growth as discussed theoretically by Blackburn and H ung (1 998), we initially intended to do as Siddiki (2002), who includes two variables for both trade liberalization and financial developm ent in the sam e regression. D ue to the existing high correlation (r = 0,98) am ong trade liberalization and financial developm ent indexes, it m ay not be appropriate to include both at the sam e tim e in a regression. T herefore, we have decided to construct another index that includes both proxies for trade liberalization and financial developm ent, nam ely X /Y , M /Y , O PE N /Y , M 1 /Y , M 1 /M 2, M 2/Y and M 2Y /Y . T his new index (E L ), therefore, involves proxies for both external liberalization and financial developm ent. In a narrow sense, this index (E L ) can be considered as an econom ic liberalization index, which carries instrum ents from both aspects of the issue concerned here. T he E L index is as follows: EL=-0,979L(M1/M2)-0,904L(M1/Y)+0,203L(M2/Y)+0,903L(M2Y/Y) +0,962L(M/Y)+0,97L(X/Y)+0,98L(OPEN) (1 1 ) opt = i =1 T he coefficients of the E L ind ex are consistent with the ab ove find ings. i it (8) 5. Em pirical Results 5.1. D ata set and Time Series Properties of the Variables T he research p eriod is d eterm ined b y the d ata availab ility. T he annual d ata is em p loyed for the T urkish econom y for the p eriod 1 963-2005. T he gross national p rod uct (G N P ) at 1 987 constant p rices is availab le from the web site of the S tate P lanning O rganization. N arrow m oney (M 1 ) and b road m oney (M 2) are taken from the web site of the International F inancial S tatistics (IF S ). M 2Y is collected from the electronic d ata d issem ination system in the C entral B ank in T urkey. T rad e variab les (M and X ) are also taken from the IF S and converted into national currency b y using the exchange rate availab le in the IF S . R ecent d evelop m ents in econom etrics req uires that b efore und ertaking an em p irical analysis, tim e series p rop erties of the d ata in term s of unit root should b e investigated b ecause regression analysis carried out with non-stationary variab les m ay invalid ate m any of the assum p tions of regression analysis. If a tim e series has a unit root, a wid esp read and convenient way to rem ove non-stationarity is b y taking first d ifferences of the relevant variab le. A non-stationary series, which b y d ifferencing d tim es transfers to a stationary one, is called an integrated of ord er d and d enoted as I(d ) (C harem za and D ead m an, 1 997 ). In fact, when a series is integrated of ord er one it m eans that it is not itself stationary, b ut that its first d ifferences are stationary. T he d efinition of the variab les is p resented in T ab le 1 . where opt rep resents the one d im ensional m easure of trad e lib eralization (or financial d evelop m ent) at tim e t, zit is the stand ard ized i th trad e lib eralization (or financial d evelop m ent) p roxy at tim e t, and i is the eigenvector com p onent that corresp ond s to a com p lem entary m easure of i th p roxy. F or trad e lib eralization, three p roxies, nam ely ratio of exp ort to incom e (X /Y ), ratio of im p ort to incom e (M /Y ) and ratio of exp ort p lus im p orts to incom e (O P E N ) are used to ob tain a trad e lib eralization ind ex (T L ): 7 TL=0,9852L(M/Y)+0,9991L(OPEN)+0,98L(X/Y) (9) where L d enotes the logarithm of the following variab les. A s can b e seen from the coefficients of the trad e lib eralization p roxies, they have p ositive im p acts on the trad e lib eralization ind ex. T he ind ex for financial d evelop m ent (F D ) includ es the m onetary aggregates, nam ely M 1 /Y , M 1 /M 2, M 2/Y and M 2Y /Y . T he F D ind ex as follows: FD=-0,905L(M1/Y)+0,23L(M2/Y)+0,941L(M2Y/Y)-0,989L(M1/M2) (1 0) where the all the letters are defined as above. T he coefficient for financial developm ent index indicates that M 1 /Y and M 1 /M 2 are negatively related to the index and the others vice versa. 7 T he results of the construction of the ind exes are availab le up on req uest. N am e of the variab le LPRY LK L SE C TL FD EL D efinition L og of p er cap ita reel incom e L og of gross fixed cap ital form ation as a p roxy for cap ital stock L og of second ary school enrolm ent rate T rad e lib eralization ind ex F inancial d evelop m ent ind ex E conom ic lib eralization ind ex T he results of D F and A D F unit root tests show that the levels of the variab les are not stationary, b ut that their first d ifferences are stationary, with or without the inclusion of a d eterm inistic trend . 5.2. Estim ation of the Long-term Production Function T he next step , in the light of `new' growth theory, is to exam ine the m ultivariate cointegration issue am ong the variab les consid ered . A ccord ingly, a m easure of p hysical cap ital (i.e. gross d om estic cap ital form ation), a m easure of hum an cap ital (i.e. the second ary school enrolm ent rate d efined as the num b er of the stud ents in the second ary school d ivid ed b y total p op ulation) and one of the ind exes, nam ely T L , F D and E L , constructed ab ove will b e includ ed in the em p irical analysis. F ollowing R oub ini and S ala-i M artin (1 992), C oe and M oghad am (1 993), P iazola (1 995), L eigh (1 996), O d ed okun (1 996, 1 999), G hatak, M ilner and U tkulu (1 995, 1 997 ), S id d iki (2002) and L iand and T eng (2006), the augm ented p rod uction function with trad e lib eralization and financial d evelop m ent can b e written as follows: T able 1 . D efinition of the V ariables T he results of the D ickey-F uller (D F ) and A ugm ented D ickey ­ F uller (A D F ) unit root tests for the variab les are p resented in T ab le 2. T he critical values are p rod uced b y the E views 5.0 econom etrics p rogram , which is b ased on the resp onse surfaces in M acK innon (1 991 ). L and D resp ectively d enote the logarithm and d ifference of the variab le in concern. V ariab les LPRY LK L SE C FD TL EL D LPRY D LK D L SE C D FD D TL D EL Intercep t without trend -0.7 50 (0) -1 .090 (0) -1 .97 1 (4) -0.452 (0) -0.293 (0) -0.1 1 0 (0) -7 .035 (0) -5.60 (0) -3.07 (3) -6.230 (0) -5.1 52 (0) -5.22 (0) 5% C ritical V alue -2.93 -2.93 -2.94 -2.93 -2.93 -2.93 -2.93 -2.93 -2.94 -2.93 -2.93 -2.93 Intercep t with trend -2.81 2 (0) -2.31 3 (0) -2.1 23 (0) -3.451 (4) -3.026 (0) -3.494 (4) -6.953 (0) -5.590 (0) -3.1 20 (3) -4.557 (7 ) -5.097 (0) -5.1 85 (0) 5% C ritical V alue -3.52 -3.52 -3.52 -3.53 -3.52 -3.53 -3.52 -3.52 -3.53 -3.54 -3.52 -3.52 R esults Y = f (K , H , X i ) f1 , f 2 , f 3 > 0 (7 ) N ot I(0) N ot I(0) N ot I(0) N ot I(0) N ot I(0) N ot I(0) I(0) I(0) I(0) I(0) I(0) I(0) W here Y is outp ut, K is p hysical cap ital, H is hum an cap ital, X i (i = 1,2,3) d enotes the trad e lib eralization, financial d evelop m ent and the joint im p act of these two com p onents of econom ic lib eralization. T he coefficients for these three variab les are the concerns of this p ap er and it is exp ected that they have p ositive im p acts on econom ic growth. T he rationality of the variab les in the p rod uction function need s som e d etail. T hose variab les exp lained in the p revious section will b e referred to in ord er to avoid rep eating them . S ince it is highly unrealistic to assum e that financial d evelop m ent is the sole or even the m ain d riving force b ehind the growth p rocess, the p otential effects of m ore conventional factors should b e isolated b y includ ing som e variab les. R enelt (1 991 ) argues that it is p ossib le to find a significant relationship b etween growth and m any of the other variab les in the em p irical literature, p articularly in cross-country stud ies. F ollowing R enelt (1 991 ), the p rod uction function includ es sup p ly sid e variab les, nam ely p hysical cap ital, lab our, hum an cap ital, trad e lib eralization and financial d evelop m ent. In ord er to und erstand the true interactions, p otential p olicy variab les are exp licitly m od elled . A ccord ing to S cott (1 992), p hysical cap ital seem s to b e a m uch m ore im p ortant d eterm inant of econom ic growth than neo-classical growth theory suggests. A n increase in investm ent not only raises the rate of econom ic growth, b ut also creates large p ositive external effects (learning effects). P hysical cap ital is ap p roxim ated b y the gross N ot: T he ord er of augm entation in the D ickey-F uller regressions is chosen using the A kaike Inform ation C riterion and the num b ers given in the b rackets in colum ns two and three rep resent the ord er of augm entation. T able 2. D F and A D F T ests for U nit Root fixed cap ital form ation in the em p irical stud ies (F osu, 1 990; G hatak, M ilner and U tkulu, 1 995; P iazolo, 1 995; M ost and van d en B erg, 1 996; A lexand er, 1 997 ; G hura, 1 997 ). T he m ain reason why cap ital form ation is not em p loyed is that there are no annual d ep reciation rates at hand . It is exp ected that the p hysical cap ital should have a p ositive effect on econom ic growth. A sub stantial b od y of recent econom ic theory has em p hasized hum an cap ital as a d eterm inant of econom ic growth. N ot only d oes the new growth theory stress the im p ortance of hum an cap ital (R om er, 1 986; L ucas, 1 988; B arro, 1 991 and 1 998; P iazolo, 1 995; G lom m and R avikum ar, 1 997 ; H wang , 1 998), b ut hum an cap ital augm ented neoclassical m od els as well (M ankiw et. al., 1 992; G ram m y and A ssane, 1 996) . T he ed ucational level of a society serves as a p roxy for the d evelop m ent of hum an cap ital. E d ucation increases the q uality of the lab our force, and therefore, the long-term p rod uction p ossib ility curve shifts outward . A n increase in the ed ucational level has a p ositive effect on econom ic growth. In other word s, a b etter ed ucated lab our force will b e m ore p rod uctive on the job b y req uiring less sup ervision and p ossessing greater initiative in hand ling job -related p rob lem s. T he p rod uction function is estim ated using the Johansen cointegration p roced ure. In p articular, econom ic theory often suggests that the p ath of certain p airs of variab les should not d iverge, at least in the long term , though they m ay d iverge in the short term d ue to seasonal factors. If the variab les continue to d iverge, m arket forces or other instrum ents com m ence to cause them to converge again. In this vein, cointegration m eans that one or m ore linear com b inations of these variab les are stationary even though ind ivid ually they are not. If these variab les are cointegrated , they cannot m ove "too far" away from each other. In other word s, if there is a long-term relationship b etween two or m ore non-stationary variab les, the id ea is that d eviations from this long term p ath are stationary (C harem za and D ead m an, 1 997 ). In contrast, a lack of cointegration suggests that such variab les have no link; they can wand er arb itrarily far away from each other. T he ap p lication of the Johansen cointegration p roced ure (Johansen, 1 988; Johansen and Juselius, 1 992) in the em p irical literature is very com m on. T his m ethod ology em p hasizes the id entification of long-term relationship s, and hence is p articularly ap p rop riate for stud ying the d eterm inants of p otential outp ut. T hese new d evelop m ents in tim e series econom etrics have b een recently exp loited in the growth literature (S erletis, 1 994; Jones, 1 995; P iazolo, 1 995; L eigh, 1 996; A restis and D em etriad es, 1 997 ; C ellini, 1 997 ; H ansson and Jonung, 1 997 ; L au and S in, 1 997 ; B atina, 1 998; H wang , 1 998; R ousseau and W achtel, 1 998; S id d iki, 2002; L iang and T eng, 2006). S tud ies m entioned ab ove utilizing the tim e series p rop erties and cointegration analysis have attem p ted to estab lish a long term relationship b etween the level of the set of the variab les in their em p irical analysis. T he Johansen cointegration p roced ure involves estim ating a V ector A utoregressive M od el (V A R ) such as (H old en and T hom p son, 1 992; C harem za and D ead m an, 1 997 ): Z t = A1 Z t -1 + ... + Ak Z t -k + Dt + ut (1 3) where Z t is a (n x 1 ) vector that contains current and lagged values of n variab les which are each assum ed to b e I(1 ), each A i is an (n x n) m atrix of p aram eters, D t is a vector of I(0) variab les8 and is the vector of rand om errors. H ere, the form ulation of the V A R m od el is of m ajor im p ortance b ecause the results of the cointegration test can b e very sensitive to that form ulation. T here are two m ain interrelated issues that p articularly should b e taken into consid eration. T he first one is to includ e an ap p rop riate lag length to ensure that the resid uals are white noise. T he second is that using too m any lags red uces the p ower of the statistics. T herefore, the choice of the ap p rop riate laglength is im p ortant. T here are several criteria to d eterm ine the ap p rop riate lag-length in the em p irical literature, nam ely the A kaike Inform ation C riteria, the S chwarz and H annan-Q uinn criteria. F ollowing the econom ic d iscussions in the p revious section, the variab les are classified und er three group s. T he first one assesses the im p acts of trad e lib eralization on the p rod uction. T he second one concerns the relationship b etween the financial d evelop m ent and incom e. F inally, the last asserts the joint effect of trad e lib eralization and financial d evelop m ent (econom ic lib eralization) on econom ic growth. T he correlation coefficient b etween trad e lib eralization and financial d evelop m ent ind exes are q uite high (r = 0,98), and therefore two variab les are not includ ed in the sam e regression. F orm ally, the three group s of the variab les are nam ed as m od els and form ed as in T ab le 3: M od el I M od el II M od el III L PRY , L K, L SE C , T L L PRY , L K, L SE C , F D L PRY , L K, L SE C , E L T able 3. D efinition of the V ariables and the Systems E m p irical investigation (carried out E -views econom etric software p rogram m e) starts from an augm ented V A R with four lags on all variab les. T he S chwarz and H annanQ uin criteria showed that, in all m od els, ap p rop riate lag length is eq ual to one (availab le up on req uest). A fter the estab lishm ent and estim ation of the V A R , the cointegration statistics, nam ely m axim um eigenvalue and trace statistics, 8 D t actually rep resents a vector of any variab les that are includ ed in the system to ensure that errors ut are white noise; thus it m ay contain d um m y variab les as well. d evelop ed b y Johansen are ap p lied to test whether there is a the long-term relationship am ong the variab les. T he results are p resented in T ab le 4. H o: rank= r Model I r= = 0 r< = 1 r< = 2 r< = 3 Model II r= = 0 r< = 1 r< = 2 r< = 3 Model III r= = 0 r< = 1 r< = 2 r< = 3 M ax E igen. 35,629* 1 9,21 3 9,242 4,444 38,260* 20,457 1 2,456 3,250 37 ,205* 1 9,61 5 8,628 4,084 5% 28,588 22,299 1 5,892 9,1 64 28,588 22,299 1 5,892 9,1 64 28,588 22,299 1 5,892 9,1 64 T race 68,529* 32,899 1 3,686 4,444 7 4,425* 36,1 64* 1 5,7 07 3,250 69,534* 32,328 1 2,7 1 2 4,084 5% 54,07 9 35,1 92 20,261 9,1 64 54,07 9 35,1 92 20,261 9,1 64 54,07 9 35,1 92 20,261 9,1 64 (E q u atio n 1 4 ). In ad d itio n , fin an cial d ev elo p m en t h as p o sitiv ely co n trib u ted to eco n o m ic g ro w th (eq u atio n 1 5 ). F in ally, alth o u g h th e m ag n itu d e o f its co efficien t is v ery sm all, th e v ariab le fo r eco n o m ic lib eraliz atio n (E L ), w h ich cap tu res th e asp ects o f b o th trad e lib eraliz atio n an d fin an cial d ev elo p m en t, also h as a p o sitiv e effect o n th e g ro w th rate. H o w ev er, p h ysical cap ital in M o d els II an d III (L K in E q u atio n 1 5 an d 1 6 ) seem s to b e in sig n ifican t. T h e em p irical an alysis sh o w s th at fin an cial d ev elo p m en t (F D ) an d trad e lib eraliz atio n (T L ) sep arately h av e p o sitiv e co n trib u tio n s to eco n o m ic g ro w th in T u rkey. In ad d itio n , th e jo in t im p act o f trad e lib eraliz atio n an d fin an cial d ev elo p m en t in term s o f eco n o m ic lib eraliz atio n (E L ) also p o sitiv ely in flu en ce eco n o m ic g ro w th . A s arg u ed b y th e en d o g en o u s g ro w th literatu re, h u m an cap ital also p o sitiv ely affects in co m e. 6. Con clusion W ith th e em erg en ce o f th e n ew g ro w th th eo ries, th e im p act o f eco n o m ic p o licy o n eco n o m ic g ro w th h as b een a h o tly d eb ated issu e. In o th er w o rd s, w h eth er eco n o m ic p o lices in a co u n try in flu en ce th e eco n o m ic g ro w th is an em p irical issu e. T h is p ap er em p irically in v estig ated w h eth er trad e lib eraliz atio n an d fin an cial lib eraliz atio n h av e h ad an y sig n ifican t im p act o n eco n o m ic g ro w th in T u rkey w h ich , as a d ev elo p in g eco n o m y, h as w itn essed an u n p reced en ted stag ed refo rm attem p t in v o lv in g extern al (trad e) an d in tern al (fin an cial) lib eraliz atio n , esp ecially after th e 2 4 J an u ary D ecisio n s fo llo w in g th e eco n o m ic crisis in 1 9 8 0 . In T u rkey, eco n o m ic lib eraliz atio n in term s o f trad e an d th e fin an cial secto r w as at th e h eart o f th e stab iliz atio n p ro g ram m e em p lo yed in 1 9 8 0 an d h as co n stitu ted an in teg ral p art o f th e eco n o m ic p o licies sin ce th en . T h e th eo retical co re o f th e eco n o m ic refo rm p ro g ram m e is to p ro v id e efficien cy in th e allo catio n o f scarce reso u rces an d to m o b iliz e u n p ro d u ctiv e reso u rces in to in v estm en t an d h en ce p ro m o te eco n o m ic g ro w th . T o test th e im p act o f trad e lib eraliz atio n an d fin an cial d ev elo p m en t o n T u rkish eco n o m ic g ro w th , th ree altern ativ e m easu res (in d exes) w ere d ev elo p ed b y m akin g u se o f th e p rin cip al co m p o n en ts an alysis, n am ely trad e lib eraliz atio n , fin an cial d ev elo p m en t an d , in a sen se, eco n o m ic lib eraliz atio n p ro xies. T h e em p irical resu lts, o b tain ed b y em p lo yin g th e m eth o d s o f th e tim e series eco n o m etrics fo r th e p erio d 1 9 6 3 2 0 0 5 , sh o w ed th at trad e lib eraliz atio n an d fin an cial d ev elo p m en t p o sitiv ely co n trib u tes to eco n o m ic g ro w th . F u rth erm o re, th e jo in t im p act o f trad e lib eraliz atio n an d fin an cial d ev elo p m en t in term s o f eco n o m ic lib eraliz atio n o n eco n o m ic g ro w th is also sig n ifican t in T u rkey. T able 4. Johansen Cointegration T est Results B oth trace and m axim um eigenvalue statistics showed that there is at least one cointegrating vector for each m od el. T race statistics ind icated that there are two cointegrating vectors in M od el II. F or the sam e m od el, the m axim um eigenvalue statistic resulted in one cointegrating vector. T herefore, one cointegrating vector for the M od el II is assum ed in the further analysis. T he corresp ond ing cointegrating vectors for each m od el resp ectively p resented as follows (stand ard errors are given in the p arenthesis): (1 4) (1 5) (1 6) A s far as th e em p irical resu lts are co n cern ed , seco n d ary sch o o l en ro lm en t rate, fin an cial d ev elo p m en t an d trad e lib eraliz atio n are statistically sig n ifican t. T h e sig n o f th e v ariab les co n cern ed (T L , F D an d E L ) is as arg u ed in th e th eo retical d iscu ssio n ab o v e. P articu larly, h u m an cap ital accu m u latio n (L S E C ) in th ree m o d els (E q u atio n 1 4 , 1 5 , 1 6 ) is sig n ifican t an d in flu en ces eco n o m ic g ro w th p o sitiv ely. T h e im p act o f trad e lib eraliz atio n (T L ) o n p er cap ita in co m e is p o sitiv e

Journal

South East European Journal of Economics and Businessde Gruyter

Published: Nov 1, 2008

References