Volume 30, Issue 1 Internatonal Captal Moblty n Afrcan Countres: Do the legal orgns matter? Chrysost Bangaké LEO, Unversty of Orléans (France) Jude Eggoh LEO, Unversty of Orléans (France) Abstract Ths paper nvestgates the Feldsten-Horoka coeffcents and legal orgns for 37 Afrcan countres usng the recently developed panel contegraton technques. The emprcal fndngs reported n the paper reveal that savngs and nvestment are nonstatonary and contegrated seres. The estmated coeffcents usng DOLS s 0.58 for the sample as a whole for the perod 1970-2006. However, there are marked dfferences n retentons ratos n each country group wth rato lowest n common law countres (0.34) compared to French cvl law countres (0.85). These results mply that n the countres wth strong legal protectons of nvestors, captal tends to be moble nternatonally than n countres wth worse protecton. We are grateful to Patrck Vlleu and Jean-Paul Polln for helpful comments. Ctaton: Chrysost Bangaké and Jude Eggoh, (2010) ''Internatonal Captal Moblty n Afrcan Countres: Do the legal orgns matter?'', Economcs Bulletn, Vol. 30 no.1 pp. 73-83. Submtted: ov 25 2009. Publshed: January 06, 2010.
1. Introducton The degree of nternatonal captal moblty determnes the effcency of captal allocaton n the world economy. It s generally admtted that most of the Afrcan countres keep sgnfcant legal restrctons over captal movements and have lmted fnancal market lnkages wth world economy. Such a stuaton has contrbuted to weak economc growth, a relatvely low savng rate and captal flght from the regon (Coller and Gunnng, 1999). Therefore, understandng the extent to whch domestc savng fnances domestc nvestment n Afrca s an mportant aspect for economc polcy makers and frms. There are several ways to nvestgate the degree of nternatonal captal moblty 1. One test proposed by Feldsten and Horoka (1980, hereafter FH) for captal moblty s to examne the relatonshp between savng and nvestment. Accordng to these authors, n the absence of captal moblty, domestc savng and nvestment are hghly correlated snce nvestment s fnanced by domestc savng. Snce the work of FH, many economsts have studed the relatonshp between savngs and nvestment 2. Ths paper nvestgates the relatonshp between the savng-nvestment takng nto account legal orgns n Afrcan countres. In the last decade, economsts have produced a consderable body of research suggestng a strong lnk between fnance and law. The dea s that a varety of legal rules (e.g., those governng both nvestor protecton and legal procedure) can nfluence the protecton of outsde nvestors and hence fnancal markets. For example, La Porta and al. (1998) show that Englsh common law countres have generally the strongest legal protectons of nvestors whle the cvl law countres the weakest. Shlefer and Wolfenzon (2002) show that n countres wth better nvestor protecton, a larger fracton of the nvested captal comes from the external market and a small fracton from nternal funds (the funds of the entrepreneurs settng up). Subsequent research showed that the nfluence of legal orgns on law and regulatons s not restrcted to fnance (see Djankov et al. 2003, for example). Unfortunately, there has only been one paper that has examned the Feldsten- Horoka puzzle and legal orgns. Gunj (2003) ntroduces a proxy of legal protecton of nvestors, a dummy varable that ndcate legal orgns, nto the Feldsten Horoka savngnvestment regresson for a sample of 20 OECD countres n 1970-2000. The estmaton shows that the French-cvl-law countres, whch have the weakest nvestor protecton, the domestc nvestment rates are generally less strongly correlated wth the domestc savng rates. Although the Gunj s study s nterestng, t focuses however on cross secton data. Savng and nvestment rates usually turn out to be nonstatonary. In the unt-root lterature, t s argued that the wdespread falure of hypothess testng n relatvely short seres may be accounted for by the low power of conventonal unvarate unt root tests aganst persstent alternatves, typcally for sample szes that occur n practce. Further, the tradtonal contegraton technque has also the problem of low power. In order to deal wth these problems, we use recently developed panel contegraton technque and dynamc OLS (hereafter DOLS) methods n order to deal wth heterogenety problems and to conduct plausble tests. To the best of our knowledge, ths s the frst paper on FH usng panel contegraton and legal orgns n Afrcan context. 1 The presence of captal moblty s tested alternatvely by usng the savng correlaton, uncovered nterest party condton, and fnally the consumpton smoothng approach to the current account. See Obstfeld (1993) for more detal. 2 See Coakley and al. (1998) for a survey. 1
The paper s organzed as follows. The next secton ntroduces the emprcal methodology. The data and emprcal results are presented and nterpreted n secton 3. Secton 4 makes some concludng remarks. 2. Emprcal methodology Followng Feldsten and Horoka (1980) we propose assessng the degree of captal moblty by analyzng the relaton between savng and nvestment. I S, (1) Y Y I S where and are respectvely the savng and nvestment rates of country, s the Y Y savngs-retenton coeffcent and s the error term. For a small, open economy where captal s perfectly moble nternatonally, should be close to zero. If equals zero, then there s no relatonshp between savng and nvestment. Feldsten and Horoka (1980) suppose that f s large, however, captal s consdered mmoble nternatonally. For example, f equals one, then all addtonal savng goes to fnance domestc nvestment. Most wrters dentfy two man secular legal tradtons: common law and cvl law, and several sub-tradtons- French, German, socalst and Scandnavan (see Laporta et al. 2008). In ths study the commercal law s classfed nto two orgns: Common law and French cvl law because of our sample. Followng Shlefer and Wolfenzon (2002), our man hypothess s that captal s relatvely moble n countres wth stronger protecton whereas n countres wth worse protecton domestc nvestment rates and domestc savng rates are strongly assocated. Our methodology departs from Gunj (2003). We make use of the new development n panel unt root tests and contegraton technques to nvestgate the relatonshp between savngs and nvestment n Afrcan countres for the perod 1970-2006. We then break up our sample nto French cvl law and common law countres and examne whether, the savng rato have change over tme. 2.1 Panel unt root tests Before proceedng to contegraton technques, we need to verfy that all varables are ntegrated to the same order. In dong so, we have used frst generaton tests of panel unt root due to Im, Pesaran and Shn (2003) and Maddala and Wu (1999) 3 and second generaton test of panel unt root of Pesaran (2005). These tests are less restrctve and more powerful compared to the tests developed by Levn and Ln (1993, 2002) 4, whch don t allow for heterogenety n the autoregressve coeffcent. The tests proposed by IPS permt to solve Levn and Ln s seral correlaton problem by assumng heterogenety between unts n a 3 Henceforth, IPS for Im, Pesaran and Shn, and MW for Maddala and Wu. 4 For a useful survey on panel unt root tests, see Hurln and Mgnon (2005) and Banerjee (1999). 2
dynamc panel framework. The basc equaton for the panel unt root tests for IPS s as follows: y p y y ; 1,2,..., ; t 1,2,...,, (2), t, t1 j, t j, t T j1 where y, t stands for each varable under consderaton n our model, s the ndvdual fxed effect and p s selected to make the resduals uncorrelated over tme. The null hypothess s that 0 for all versus the alternatve hypothess s that 0 for some 1,..., 1 and 0 for 1 1,...,. The IPS statstc s based on averagng ndvdual ADF statstcs and can be wrtten as follows: 1 t ( tt ), (3) 1 where t T s the ADF t-statstc for country based on the country-specfc ADF regresson, as n Eq (1). IPS show that under the null hypothess of non statonary, the t statstc follows the standard normal dstrbuton asymptotcally. The standardzed statstc t IPS s expressed as: 1 n t EtT 0 1 tips. (4) 1 Var tt 0 1 Maddala and Wu (1999) argue that whle Im et al. s tests relax the assumpton of homogenety of the root across the unts, several dffcultes stll reman. They suggest the use of a Fsher type test whch s based on combnng the p-values, of the test-statstc for a unt root n each cross-sectonal unt. The MW test statstc s gven by: 2 ln. (5) 1 The MW test statstc s dstrbuted as Ch square wth 2 degrees of freedom under the hypothess of cross-sectonal ndependence. Accordng to Bretung (1999), IPS s test s not powerful when ndvdual trends are ncluded. Ths test s senstve to the specfcaton of determnstc trends compared to IPS s test. The MW test has the advantage that ts value does not depend on dfferent lag lengths n the ndvdual ADF regressons. Furthermore, Maddala and Wu (1999) found that MW s test s superor compared to IPS s test. Both the tests (IPS and MW) have the drawback to suppose that the cross-sectons are ndependent; the same assumpton s made n all frst generaton of panel unt root. However, t has been ponted out n the lterature that cross secton dependence arses due to unobserved common factors, externaltes, regonal and macroeconomc lnkages, and unaccounted resdual nterdependence. Recently, some new panel unt root tests have emerged and address the queston of the dependence and correlaton gven the prevalence of macroeconomc dynamcs and lnkages. These tests are called the second generaton panel unt root tests. The well-known second generaton test that s consdered n ths paper s the Pesaran s CIPS test (2005). In order to formulate a panel unt root test wth cross-sectonal dependence, Pesaran (2005) consders the followng Cross-Sectonal Augmented Dckey-Fuller (CADF) regresson, estmated the OLS method for the th cross-secton n the panel: 3
y y y y y k t, t1 t1 j, t j, t j t j0 j0 1 1 yt y t yt yt t 1 1 k (6) where, 1, 1,, T s the t-statstc of the estmate of n the above equaton used for computng the ndvdual ADF statstcs. More precously, Pesaran proposed the followng test CIPS statstc that s based on the average of ndvdual CADF statstcs as follows: 1 CIPS t, T. (7) 1 The crtcal values for CIPS for varous determnstc terms are tabulated by Pesaran (2005)., and 2.2 Panel contegraton tests Once the order of statonary has been defned, we would apply Predron s contegraton test methodology. Indeed, lke the IPS and MW panel unt root, the panel contegraton tests proposed by Pedron (1999) also take n account heterogenety by usng specfc parameters whch are allowed to vary across ndvdual members of the sample. Takng nto account such heterogenety consttutes an advantage because t s unrealstc to assume that the vectors of contegraton are dentcal from an ndvdual to another for the panel. The mplementaton of Pedron s contegraton test requres estmatng frst the followng long run relatonshp: yt t 1 x1, t 2 x2, t... M xm, t t (8) for 1,..., ; t 1,..., T ; m 1,..., M where refers to the numbers of ndvdual members n the panel; T refers to the number of observaton over tme; M refers to the number of exogenous varables. The structure of estmated resduals s follows: ˆ ˆ uˆ. (9) ˆt t1 t Pedron has proposed seven dfferent statstcs to test panel data contegraton. Out of these seven statstcs, four are based on poolng, what s referred to as the Wthn dmenson and the last three are based on the Between dmenson. Both knds of tests focus on the null hypothess of no contegraton. However the dstncton comes from the specfcaton of the alternatve hypothess. For the tests based on Wthn, the alternatve hypothess s 1 for all, whle concernng the last three test statstcs whch are based on the Between dmenson, the alternatve hypothess s 1, for all. The fnte sample dstrbuton for the seven statstcs has been tabulated by Pedron va Monte Carlo smulatons. The calculated statstc tests must be smaller than the tabulated crtcal value to reject the null hypothess of absence of contegraton. 2.3 Panel contegraton estmaton Although Pedron s methodology allows us to test the presence of contegraton, t could not provde estmaton of long-run relatonshp. For panel framework, n presence of contegraton, 4
several estmators are proposed: OLS, Fully Modfed OLS (FMOLS) 5, and dynamc OLS (DOLS). In ths paper, we use the dynamc ordnary least squares (DOLS) developed by Kao and Chang (2001) to estmate the long-run vector between savng and nvestment. Ths estmator s desgned for non statonary panels and corrects the standard pooled OLS for seral correlaton and endogenety of regressors that are normally present n long run economc relatonshps. Furthermore, Kao and Chang (2001) showed that both the OLS and FMOLS exhbt small sample bas and the DOLS estmator appears to outperform both estmators. The DOLS s an extenson of Stock and Watson s (1993) estmator. In order to obtan an unbased estmator of the long-run parameters, DOLS estmator uses parametrc adjustment to the errors by ncludng the past and the future values of the dfferenced I(1) regressors. The dynamc OLS estmator s obtaned from the followng equaton: jq t t j t j t jq y x c x v 2,. (10) 1 where c j s the coeffcent of a lead or lag of frst dfferenced explanatory varables. The estmated coeffcent of DOLS s gven by: T 1 T ˆ DOLS zt z t zt yt 1 t1 t1 ˆ where zt xt x, x, tq,..., x, tq s 2q 1 1 vector of regressors. (11) 3. Data and emprcal results 3.1. Data The data are taken from the World Development Indcators (WDI, 2008) CD-ROM for 37 Afrcan countres for the perod 1970-2006. Followng the orgnal study of Feldsten and Horoka, savngs s defned as gross domestc savngs as a percentage of GDP whle nvestment s measured by gross fxed captal formaton dvded by GDP 6. The data are summarsed n table 1 (n appendx) whch shows marked dfferences between savngs and nvestment ratos wthn and across countres. 3.2. The unt root tests Table 2 reports the outcome for the global sample of three panel unt root tests: IPS (2003), MW (1999) and Pesaran (2003). It shows that the null hypothess of the unt roots for the panel data for the nvestment and savngs seres cannot be rejected n level. However, ths hypothess s rejected when seres are n frst dfferences. These results strongly ndcate that the varables n level are non-statonary and statonary n frst-dfferences. The same ssues 5 The FMOLS s popular n conventonal tme seres econometrcs, for t s beleved to elmnate endogenety n the regressors and seral correlaton n the errors. 6 Bayoum (1990) and Snha and Snha (2004) suggest the use of gross fxed captal formaton, snce t excludes the procyclcal nventores component that may lead to spurous correlatons wth savngs. 5
are obtaned for the panels of country groups: cvl law and common law countres (table 3) Therefore, we can mplement a test for panel contegraton between savngs and nvestment. Table 2: Panel unt root for savng and nvestment ratos, 1970-2006 (global sample) Im et al. (2003) Maddala and Wu (1999) Pesaran (2005) Statstc P-values Statstc P-values Statstc P-values All countres (37) I/ Y -1.547 0.148 59.477 0.890-1.421 0.988 t t I/ Y -4.026 0.000 488.481 0.000-3.913 0.000 t S/ Y -1.384 0.489 63.166 0.811-1.599 0.866 S/ Y -4.149 0.000 544.287 0.000-3.766 0.000 t Table 3: Panel unt root for savng and nvestment ratos, 1970-2006 (cvl law vs common law countres) Im et al. (2003) Maddala and Wu (1999) Pesaran (2005) Statstc P-values Statstc P-values Statstc P-values Cvl law countres I/ Y -1.480 0.322 35.583 0.669-1.610 0.776 t t I/ Y -4.052 0.000 273.646 0.000-3.948 0.000 t S/ Y -1.244 0.735 24.456 0.974-1.434 0.945 S/ Y -4.300 0.000 320.241 0.000-3.904 0.000 t Common law countres I/ Y -1.671 0.115 19.893 0.953-1.279 0.982 t t I/ Y -3.876 0.000 191.575 0.000-3.871 0.000 t S/ Y -1.509 0.297 29.348 0.601-1.081 0.998 S/ Y -3.986 0.000 211.272 0.000-3.792 0.000 t 3.3. Panel contegraton tests Table 4 shows the outcomes of Pedron s (1999) contegraton tests between the nvestment and savngs rates. We use four wthn-group tests and three between-group tests to check whether the panel data are contegrated. The columns labeled wthn-dmenson contan the computed value of the statstcs based on estmators that pool the autoregressve coeffcent across dfferent countres for the unt root tests on the estmated resduals. The columns labeled between-dmenson report the computed value of the statstcs based on estmators that average ndvdually estmated coeffcents for each country. Except the v-statstc test, the results of the wthn-group test and the between-group tests show that the null hypothess of no contegraton can not be rejected at the 1% sgnfcant level. Therefore, the ratos of savng and nvestment are contegrated for the panel of all countres and for the panels of country groups (common law and cvl law countres) The presence of a long-run relatonshp between nvestment and savng rate n the panel of Afrcan countres are economcally meanngful n that t suggests that these countres meet the long-run solvency condton. Havng found that there exsts a contegratng lnk between the two varables (ratos of savng and nvestment), t s convenent that the savngs-retenton 6
coeffcent be estmated usng a panel contegratng estmator. In ths paper, we choose to employ the Dynamc OLS (DOLS). Table 4: Pedron Panel contegraton test results, 1970-2006 Wthn-dmenson (panel) Between-dmenson (group) v -Stat -Stat PP-Stat ADF-Stat -Stat PP-Stat ADF-Stat All 0.538-4.836 *** -6.282 *** -7.282 *** -2.659 *** -5.438 *** -6.032 *** French 1.564 * -5.093 *** -6.274 *** -7.623 *** -2.389 *** -4.686 *** -5.835 *** Englsh -0.533-0.875-1.375 * -1.731 *** -0.415-1.995 ** -2.729 *** otes: Results wth a trend and tme-dummes. The test statstcs are normalzed so that the asymptotc dstrbuton s standard normal. *, **, *** ndcate rejecton of the null hypothess of non contegraton at the 10, 5, and 1 percent sgnfcance levels, based respectvely on crtcal values of 1.281, 1.644 and 2.326. 3.4. Panel contegraton estmatons We estmate the contegratng vector usng panel DOLS method. The estmates results by perod are reported n table 5. Frst, for the pool of all countres the FH coeffcent for the full sample perod (1970-2006) s 0.58 and broadly consstent wth the work by Adedej and Thornton (2006) whch report a savngs-retenton coeffcent of 0.51 for DOLS estmate. However, our result contrast wth those of De Wet and Van Eyden (2005) whch report a savngs retenton coeffcent small (0.31, 0.34 and 0.28) respectvely for Pooled, fxed effects and random effects models. Furthermore, we show that the FH coeffcents declnes over the sample perod. Indeed, the FH coeffcent s 0.62 for the perod 1970-1987, whereas t s 0.52 n the second sub-perod (1988-2006) mplyng that captal moblty ncreased n Afrcan countres over tme. There are strkng results when consderng panels of country groups accordng to ther legal orgns (French cvl law versus common law countres). There are marked dfferences n retentons ratos between country groups. The estmates for French cvl law countres ndcate a savng retenton coeffcent relatvely hgh, 0.85 compared to common law countres wth 0.34 for the entre perod (1970-2006). When we consder two sub-perods (1970-1987 and 1988-2006), the FH coeffcent declnes n French cvl law countres but remans relatvely hgh (from 0.98 to 0.70), whle n common law countres the FH coeffcent s broadly same about 0.30. These results ndcate that captal has been relatvely moble nternatonally n common law countres whle n cvl law countres the FH puzzle s held although the coeffcent declnes over perod. Our nterpretaton of the FH coeffcent s dfferent of Gunj (2003) because he takes nto account a dummy varable, whle our analyss centres drectly on the coeffcent of. What do the marked dfferences n retentons ratos between country groups (common law countres versus French cvl law countres) means? Accordng to La Porta and al. (1998), Englsh common law countres have generally the strongest legal protectons of nvestors whle the cvl law countres the weakest. Shlefer and Wolfenzon (2002) show that n countres wth better nvestor protecton, a larger fracton of the nvested captal comes from the external market and a small fracton from nternal funds (the funds of the entrepreneurs settng up). Ths fact combned to less formalsm of judcal procedures and many reforms undertaken early 1980 n common law countres (prvatzaton, ratonalzaton of ther publcs sector, lberalzaton of exchange rate and fnancal systems) have contrbuted to relatve captal moblty. 7
Table 5 : Panel contegraton vector: DOLS results Countres Perod Savng-retenton rato t-rato All 1970-2006 0.582 *** 12.29 1970-1987 0.620 *** 7.43 1988-2006 0.526 *** 7.48 Common law 1970-2006 0.343 *** 3.92 1970-1987 0.283 ** 2.25 1988-2006 0.335 *** 2.70 French Cvl law 1970-2006 0.859 *** 30.48 1970-1987 0.982 *** 27.77 1988-2006 0.708 *** 19.12 otes: the value n parenthess denotes the t-value for zero coeffcents. *** sgnfcant at 1% and ** sgnfcant at 5%. 4. Concludng remarks Ths paper has studed the nternatonal captal moblty of 37 Afrcan countres n terms of the FH coeffcent and legal orgns, applyng recently developed panel contegraton methods. We apply panel contegraton test and used DOLS methods n order to deal wth heterogenety problems and to conduct plausble tests. The emprcal fndngs reported n the paper reveal that savngs and nvestment are nonstatonary and contegrated seres. Captal was relatvely moble n the Afrcan countres n the sample durng 1970-2006, wth estmated savng retenton coeffcents of 0.58 for the pool of all countres. When we consder panels of country groups (French cvl law countres versus common law countres) there are marked dfferences n retentons ratos n each country group wth rato lowest n common law countres compared to French cvl law countres. Our results show that captal tends to be moble nternatonally n common law countres wth strongest legal protecton than n French cvl law wth the weakest. Some recent papers also suggest that n countres wth better nvestor protecton, a larger fracton of the nvested captal comes from the external market and a small fracton from nternal funds. References Adedej, O and J. Thornton (2006) Savng, Investment and captal moblty n Afrcan countres Journal of Afrcan Economes, 16, 393-405. Banerjee, A (1999) Panel data unt roots and contegraton: an overvew Oxford Bulletn of Economcs and Statstcs, 61, 607 629. Bayoum, T. (1990) Savng-Investment correlatons: mmoble captal government polcy or endogenous behavour? Internatonal Economc Journal, 37, 360-387 Bretung, J. (2000) The Local Power of Some Unt Root Tests for Panel Data n Advances n Econometrcs, by Baltag, 161-178. Chen, B., S. McCoskey, and C. Kao (1999) Estmaton and Inference of a Contegrated Regresson n Panel Data: A Monte Carlo Study Amercan Journal of Mathematcal and Management Scences, 19, 75-114. Coakley, J., A.M Fuertes, A.M., and Spagnolo, F (2001) The Feldsten- Horoka puzzle s not as bad as you thnk. Avalable from: http://papers.ssrn.com/sol3/papers.cfm. Coller, P. and J.W Gunnng (1999) Why has Afrca grown so slowly. Journal of Economc Perspectves, 13, 3-22. 8
De Wet, A.H and R. Van Eyden (2005) Captal moblty n Sub-Saharan Afrca: A panel data approach South Afrcan Journal of Economcs, 73, 1-22 Djankov, S, C. Mc Lesh, T. enova, and A. Shlefer. (2003) Who owns the meda? Journal of Law and Economcs, 46, 341-381. Feldsten, M. and C. Horoka (1980) Domestc savng and nternatonal captal flow. The Economc Journal, 90, 314-329. Gunj, H (2003). The Feldsten-Horoka puzzle and Law. Economcs Bulletn, 11, 1-12. Hurln, C. and V. Mgnon (2005). Une synthèse des tests de racne untare sur données de panel. Econome et Prévson, 169, 253-294. Im, K.S, M. H Pesaran and Y. C Shn (2003). Testng for unts roots n heterogeneous panels. Journal of Econometrcs, 115, 53-74. Kao, C. and M.H Chang (2000) On the estmaton and nference of a contegrated regresson n panel data. Advances n Econometrcs, 15, 179-222. La Porta R., F. Lopez-de-Slanes., A., Shlefer and R.W Vshny (1998) Law and Fnance. Journal of Poltcal Economy, 106, 1-53. La Porta R., F. Lopez-de-Slanes and A. Shlefer (2008) The Economc consequences of legal orgns. Journal of Economc Lterature, 285-332 Levn, A. and C. F Ln (1993) Unt Root Tests n Panel Data: ew Results. Dscusson paper, Department of Economcs, UC-San Dego. Levn, A., C. Ln, and C. Chu (2002) Unt root test n panel data: asymptotc and fnte sample propertes, Journal of Econometrcs, 108, 1-24. Maddala, G.S., and S. Wu (1999) A comparatve study of unt root tests wth panel data and new smple test Oxford Bulletn of Economcs and Statstcs, 61, 631-652. Obstfeld, M. (1993) Internatonal captal moblty n the 1990 BER workng paper number 4534. Payne, J.E., and R. Kumazawa (2005). Captal moblty, foregn ad, and Openness: further panel data evdence from Sub-Saharan Afrca. Journal of Economc and Fnance, 29, 122-126. Pedron, P. (1999 Crtcal values for contegraton tests n heterogeneous panels wth multple regressors. Oxford Bulletn of Economcs and Statstcs, 61, 653-670. Pesaran, M. H. (2005) A smple panel unt root test n presence of cross secton dependence Cambrdge Unversty Workng Paper n Economcs, number 346. Snha, T. and D. Snha (2004). The Mother of all puzzles would not go away Economcs Letters, 82, 259-267 Shlefer, A. and D. Wolfenzon (2002 Investor protecton and equty markets Journal of Fnancal Economcs, 106, 1-53. Stock, J. and M. Watson (1993). A smple estmator of contegratng vectors n hgher order ntegrated systems Econometrca, 66, 3-27 Appendx A. Legal orgns (Source: Laporta et al., 2008) French cvl law countres: Algera, Benn, Burkna, Cameroon, Central Afrcan Republc, Chad, Congo Democratc Republc, Congo, Côte d Ivore, Egypt, Gabon, Gunea-Bssau, Madagascar, Mal, Maurtana, Morocco, ger, Senegal, Togo, Tunsa. Common law countres: Botswana, Burund, Gamba, Ghana, Kenya, Lesotho, Malaw, gera, Rwanda, Serra Leone, South Afrca, Sudan, Swazland, Tanzana, Uganda, Zamba, Zmbabwe 9
B. Table 1: Property of the data: descrptve statstcs by country, 1970-2006 Country Investment rate Savng rate Mean Std. Dev. Mean Std. Dev. All countres 19.212 9.128 11.160 17.213 Algera 30.439 6.527 35.008 8.490 Benn 16.053 3.374 1.631 4.206 Botswana 28.227 6.778 35.846 12.467 Burkna 18.901 3.324 2.765 5.509 Burund 10.964 4.679-2.019 7.563 Cameroon 19.767 7.322 19.919 4.281 Central Afrcan Rep. 11.346 3.707 2.259 4.631 Chad 15.637 11.926 2.941 14.052 Congo. Dem. Rep. 10.770 4.985 9.675 5.308 Congo. Republc of 27.720 8.546 30.232 17.376 Cote d Ivore 15.501 6.206 21.362 5.651 Egypt 21.432 5.848 13.925 3.076 Gabon 31.905 10.131 48.911 11.673 Gamba 17.939 6.036 6.186 4.397 Ghana 14.945 7.996 7.295 3.429 Gunea-Bssau 23.953 9.399-2.492 6.810 Kenya 18.640 2.234 16.448 4.847 Lesotho 38.663 15.527-45.914 21.343 Madagascar 12.368 4.578 4.975 3.721 Malaw 17.966 5.281 9.037 6.711 Mal 19.049 4.679 4.838 6.632 Maurtana 21.932 13.424 2.713 14.470 Morocco 22.813 4.335 17.591 4.065 ger 13.344 5.163 5.679 4.711 gera 20.538 4.773 24.307 8.808 Rwanda 14.528 3.333 1.746 9.474 Senegal 18.200 4.715 7.181 3.755 Serra Leone 10.100 3.502 6.749 10.386 South Afrca 20.833 4.979 24.290 5.642 Sudan 13.545 3.920 8.485 5.087 Swazland 22.527 7.106 11.902 14.325 Tanzana 16.832 5.439 4.432 4.193 Togo 19.261 4.433 13.136 11.725 Tunsa 25.569 3.854 22.731 2.154 Uganda 13.014 5.220 5.839 4.461 Zamba 18.386 8.109 17.776 12.486 Zmbabwe 17.246 3.545 15.559 5.721 10