A model to estimate the effect of global crisis on the convergence process in EU*

Transcription

1 A model to estmate the effect of global crss on the convergence process n EU* Albu Lucan-Lvu Insttute for Economc Forecastng, Romanan Academy 711, Bucharest - Romana, Casa Academe Romane, Calea 13 Septembre Nr. 13, sector Tel. & Fax: ; E-mal: albul@pe.ro; http: Paper proposed for the EcoMod Conference, Unversty of the Azores, Ponta Delgada, Portugal, June 29, 11 - July 1, 11 Abstract There are evdences that the actual global crss affected the convergence process n EU. Generally, just new adhered countres were more affected by the actual crss. Today all forecasts are sufferng by uncertanty. There are dfferent opnons regardng how deep and how long the convergence process wll be affected. Synthetcally, the pessmstc authors are vewng the future economc dynamcs as one of so-called L type or U type or W type. Comng from lessons done by standard economc growth theores (Ramsey model, Solow-Swan model, Mankw, Romer, and Wel model, etc.) and emprcal evdences, we are consderng the convergence n the level of ncome per capta as a result of structural changes n economy. In a frst part of the study we nvestgate the dfferences among countres n EU n terms of the share n total economy of man sectors. Then, based on the spatal (emprcal) dstrbuton of such shares n EU we are proposng a model to estmate a typology of the convergence process n the European area. Takng nto account the exstng dfferences among sectors n matter of productvty, there are two versons of the model: one consderng the share of sectors n total employment and the other one n GDP. Fnally, we are usng the model to evaluate the negatve mpact of actual crss on the convergence process, how deep and how long t wll be prolonged, how nvestment and consumpton are affected. Keywords: convergence, structural changes, spatal dstrbuton, smulaton model JEL Classfcaton: C13, C31, E21, E27, O11, O47, O2 * Ths work was supported by CNCSIS project nu mber PNII IDEI 14 contract 929/8. 1

2 1. Introducton Economc theory of development postulates major changes n the structure of natonal economes along wth the hstorcal growth process. At hstorcal scale, n natonal economes frstly agrculture predomnates; then ndustry s the predomnant sector; and fnally sector of servces becomes the major part of economy. Accordng to a general rule, durng the frst stage of development, along wth a general ncrease of ncome, the demand for agrcultural goods s growng, but slower than ncome due to a smaller elastcty of ncome relatvely to ther demand. Contrary, n case of manufactured goods there wll be a larger elastcty of ncome relatvely to ther demand. Thus, the share of secondary sector n economy wll ncrease. However, n the hstorcal process of development, ncome contnung to ncrease, comng from one moment people begn to consume more servces, takng nto account that n ther case ncome elastcty relatvely to demand s even larger. Consequently, the tertary sector wll develop faster. Ths general rule s supposed to gude development at hstorcal scale, but based only on emprcal facts. Such schema may be sometmes false. Thus, may exst underdeveloped countres from the level of ncome per nhabtant vewpont, n whch tertary sector s predomnant as consequence of an extended actvty n toursm, concomtantly exstng a non-developed secondary sector. Ths stuaton mples major rsks. For nstance, n case of a deep recesson n countres supplyng toursts can strongly affect ncome from toursm n destnaton country. Further, the overall effect wll compromse at a large scale the general development process n ths country. In cases where there s not developed ether prmary sector or secondary sector to be potentally re-mproved, loan and ncreasng debt wll be alone solutons. In actual world expanson of tertary sector s concdent just wth the eme rgence and fast development of so-called new economy. Thus, the new economy s often vewed as economy of servces. Many authors consder as base of spectacular growth of tertary sector n developed countres durng last tme certan actvtes such as: scentfc research and technologcal development; desgn and expermentaton; marketng and trade (ncludng external trade); producton, stockng, processng and transmsson of nformaton; mprovement of human factor, educaton, health, and ncreasng of lfe qualty (ncludng qualty of envronment, lesure, toursm); fnancal actvty, bankng, nsurance socetes, and captal markets, etc. Just such servces, from whch essentally depends the effcency even n so-called materal sphere of producton, demonstrate today hghest dynamcs. They are ether ntegrated together wth proper productve actvtes n the same system or developed as autonomous systems, such as: bankng ndustry, toursm ndustry, nformaton ndustry, etc. 2. Theoretcal model and emprcal evdences Economc theory usually uses a number of stylsed facts of structural changes along wth economc growth process. Accordng to t, a satsfactory theory of structural changes should be able to explan the real evoluton llustrated by emprcal data. Among conclusons three stylsed facts are hghlghted: the share of prmary sector shows contnung decrease convergng on long-run to a small constant value; the share of secondary sector ncreases untl a maxmum value but further t decreases convergng on long-run to a constant value; the share of tertary sector shows contnung growth convergng on long-run to a hgh value. Consequently, a model of structural changes should be able to smulate such dynamcs. 2

3 In order to estmate parameters descrbng medum- and long-run evoluton of structure of dfferent natonal economes usually are used ether econometrc models or alternatvely they should be calbrated to ft reasonably emprcal data. To buld a theoretc model, n ths case essentally non- lnear model, we consder some lmtvalues to whch trajectores of the shares n case of the three sectors are asymptotcally convergng on long-run, functon of the level of GDP per nhabtant. The basc hypotheses, plausble from theoretcal vewpont, should be also n accordance wth emprcal data. There are three hypotheses that we used for the model, as follows: na = h = ct., for y + (1) n =, for y (2) ns = d = ct., for y + (3) where na, n and ns are shares n employment of prmary sector (manly agrculture), secondary sector (manly ndustry), and respectvely tertary sector (servces). Comng from these hypotheses dynamcs of shares of agrculture and servces n total employment can be functon of GDP per capta, y, expressed by the followng two relatons: na (y) = (A*h*y + m*b) / (A*y + m) (4) ns (y) = d / (1 + e b-c*y ) () where A, h, m, B, d, and c are calbrated parameters (they can be also econometrcally estmated); e s base of natural logarthms. Moreover, consderng the complementary relaton, na+n+ns=1, one should wrte also dynamcs of the share of ndustry n total employment: n (y) = 1 {[(A*h*y + m*b) / (A*y + m)] + [d / (1 + e b-c*y )]} () Also, takng nto account hypothess (2), we obtan the followng mplct relaton: B = 1 - [d / (1 + e b )] (7) Based on avalable cross-secton statstcal data n perod 19-, for a number of about 1 countres (ncludng all groups of countres, from poorest to rchest), and on hypothess of some long-run asymptotcal trajectores, we calbrated the model. Smulatng of the model demonstrated that n case of ndustry there are a local mnmum and a global maxmum, correspondng to two specfc crtcal values of ncome per nhabtant. Based on the model we can also smulate certan relevant long-run trajectores of structural changes. For nstance, usng the followng set of values for parameters, A=2, h=.2, m=3, d=.8, b=1.12, and c=.21, from the smulatng of the model resulted n case of ndustry a maxmum of ts share n total employment, n, equal to around 42% (correspondng to a crtcal value of GDP per capta y= USD) and a mnmum equal to around 14.7% (correspondng to y=28 USD). The complete map of smulaton s shown n Fgure 1 (where y s n thousand USD). Dscrepances among countres can be vewed now not only as dfference n ncome per nhabtant but also n terms of structural gap. Moreover, the smulatng of model demonstrates a general convergng of structures on long-run along wth economc growth process. 3

4 Fgure Spatal dstrbuton of some macroeconomc varables n Europe In context of actual convergence polcy n EU-27, t s useful to analyse the spatal dstrbuton of some basc macroeconomc ndcators. Moreover, accordng to recent avalable data from EUROSTAT for EU countres we used as output of smulaton models some sgnfcant 3D graphcal representatons and ther attached so-called geodesc maps or contour plots. Among selected macroeconomc varables, the most sgnfcant s GDP per nhabtant. In Fgure 2 s shown ts spatal dstrbuton n 7 (before global crss), as a stylsed map of EU, where LO s longtude (on ts left sde relatng to the orgn, merdan, we changed West longtude, as t s marked usually on geographcal maps, n negatve values), LA lattude, and ypps level of GDP per capta n thousand Euro PPS (Purchasng Power Standards). On the stylsed map of EU-27 we can see two dstnct groups of regons delmted by to contour lnes (red colours) and respectvely by to 1 contour lnes (blue colours) representng hghest and respectvely lowest GDP per capta levels. As two general rules, GDP per capta level s ncreasng from the rght sde of EU stylsed map (eastern EU regons) to the left sde (western EU regons) and respectvely from the bottom (southern EU regons) to the top (northern EU regons). Moreover, n Fgure 3 s shown the spatal dstrbuton of GDP per capta, as dfferences from the average EU level (EU-27 = 1) n 9 (n the mddle of global crss). Spatal dstrbutons n EU of other macroeconomc varables consdered n the convergence programme are presented n Fgure 4 nflaton, at the end of December 1 (=1), and n Fgure unemployment rate, at the end of December 1. 4

5 1 ypps ypps Fgure y_ue_ y_ue_1 Fgure 3.

6 Fgure 4. 1 u u 7 Fgure. In order to llustrate how the global crss affected the convergence n EU, we are presentng comparatvely, n Fgures and 7, the dstrbutons n EU of the GDP growth rate n the perod -7 (average annual growth rate) and respectvely n the perod 8-9 (average annual growth rate). We can see a dramatc change n GDP growth rate dstrbuton between the years before the crss and those n crss (the year 8 was excluded because n t some countres were already affected by crses but others were not yet affected). Durng last perod, the most affected countres by crss are just those regsterng lower level of GDP per capta n EU (as a rule, they are the new adhered members located n the Eastern area of

7 Europe). Thus, as the global crss wll delay the recoverng process n less developed countres of EU as much the convergence process wll be affected y% y% Fgure. 1 y% y% Fgure Contnung the convergence n UE-27 Based on the study of structural changes by stages of economc development resulted that the dfferences among countres can be evaluated by dscrepances n servces sector contrbuton both n total employment and n GDP. Analysng data on the share of servces n GDP n an 7

8 hstorcal short perod, 199-7, demonstrates a strong expandng tendency for all EU countres. Regardng ths crteron of convergence, Romana s the frst country wthn EU, regsterng an ncrease of 1.9 percentage ponts, from 38.8% to.7% (Latva, placed on the second poston, regstered n the same perod a growth of 1.7 percentage ponts, from.% to 73.3%). However, Romana contnues to be on the last place n EU regardng the share of servces n total GDP. Consequently, n case of Romana, the shares of agrculture and respectvely of ndustry n total GDP are among hghest n the EU (.% and respectvely 37.8%, n 7). In order to estmate trends n structural convergence n EU functon of economc growth we used a model just a lttle dfferent from the prevous theoretcal model. Statstcal data are referrng to 7 (thus before the crss). We calbrated econometrc model by supposng that exst certan lmt-values to whch each of the three trajectores are tendng along wth the ncome per capta growth. Thus the specfcaton of the model s n lne wth both long-run growth theory and emprcal data suppled by economc hstory. These hypotheses are synthetcally expressed by the followng equatons used for regresson n case of agrculture sector, ya, and respectvely servces sector, ys: ya_e (y) = [ (k1*y + k2) / (k3*y + k4) ] (8) ys_e (y) = [ k / (1 + k*e k7*y ) ] (9) where y s GDP per capta (we also used GDP per capta n Purchasng Power Standards), k1...k7 are estmated parameters, and e s the base of natural logarthms. In order to estmate the share of ndustry sector n GDP, y, smply we operate substtuton of the above two relatons n the balance relaton, ya+y+ys=1, obtanng the followng equaton: y (y) = 1 {[(k1*y + k2) / (k3*y + k4)] + [k / (1 + k*e k7*y )]} (1) The results of applyng the cross-secton model (usng GDP n PPS) on EU countres are presented n Fgures 8-1 (where the two dashed lnes delmt the confdence statstc nterval). Moreover, n Fgure 11 s shown the resulted general theoretcal model at the level of entre EU for 7. Thus, as mnmum for the share of agrculture sector resulted a value close to % and as maxmum for the share of servces sector resulted a value equal to about 87%. These values show that n case of new adhered countres a sgnfcant gap relatng to the average EU level n matter of structural changes stll exsts. In case of ndustral sector resulted a value of global maxmum equal to about 31.1% (correspondng to a crtcal value of GDP level per capta equal to about 1 PPS) and respectvely a value of long-run mnmum equal to about 13.4% (n case of a very large ncome per capta). More detaled nterpretaton can be extracted from so-called surface plot or 3D map and contour plot representatons of the estmated EU model (see Appendx 1). Accordng to the resulted cross-secton model (estmated on the base of 7 data) we can evaluate long-run dynamcs structural changes for each ndvdual country. Thus, the actual gap between new adhered countres and average level n EU could be nterpreted as delay n tme, ther actual structure of economy representng a smlar stuaton wth that exstng n developed western countres n EU 1- years ago. Moreover, there are evdences demonstratng that the long-run trends n new members of EU wll be smlar to those regstered n western countres. 8

9 1 y a% y a_e y a_l y a_u Fgure 8. y % y _E y _L y _U Fgure 9. 9

10 9 y s% 8 y s_e y s_l y s_u Fgure 1. 8 y at( y) y st( y) y T( y) y Fgure 11. We also appled the model of structural changes on more dgtalsed data from EUROSTAT, namely NUTS2 (comprsng around 39 regons n EU). The resulted estmatons are some dfferent but the conclusons generally are stll mantaned (see Appendx 2). Moreover, accordng to the avalable data (Nuts 2 database for 7) we analysed correlatons for a number of macroeconomc varables n case of EU-27. The selected varables are as follows: y - GDP per nhabtant n current prces (Eur); ypps - GDP per nhabtant n PPS (Eur); rac - Actvty rate (actve populaton/total populaton aged and over, %); roc - Rate of occupancy (occuped populaton/total number of populaton, %); rpop - Rate of populaton aged over (%); u - Unemployment rate (%); rp_14 - Rate of populaton aged -14 (%); 1

11 rem - Employment rate (employed populaton/total populaton aged -4, %); rp_4 - Rate of populaton aged -4 (%); ag% - Share of agrculture (plus huntng, forestry and fshng) n labour force; n% - Share of ndustry (plus constructon) n labour force; se% - Share of servces n labour force; rp_max - Rate of populaton aged and over (%); rimb - Ageng rate (populaton aged and over/ populaton aged -14, %). Usng Nuts 2 database, at the EU-27 level, GDP per capta s strongly correlated postvely wth some varables, as follows: 1) Share of servces (se%) - corr( y, se% ) =.7 2) Occupancy rate (roc) - corr( y, roc ) =.88 3) Employment rate (rem) - corr( y, rem) =. 4) Actvty rate (rac) - corr( y, rac ) =.438 ) Rate of populaton aged -14 (rp_14) - corr( y, rp_14) =.222 and negatvely correlated wth the followng varables: 1) Share of agrculture (ag%) - corr( y, ag% ) =.38 2) Share of ndustry (n%) - corr( y, n% ) =.49 3) Rate of populaton aged -4 (rp_4) - corr( y, rp_4) =.3 4) Unemployment rate (u) - corr( y, u) =.24 Between GDP and varables as Rate of populaton aged over (rpop), Rate of populaton aged and over (rp_max), and Ageng rate (rimb), there are nsgnfcant correlatons (values near ). Selected bblography Albu, L.-L. (8): Trends n Structural Changes and Convergence n EU, Romanan Journal of Economc Forecastng, 1, pages Albu, L.-L., Ghzdeanu, I., and Stanca C. (11): Spatal Dstrbuton of the Informal Economy. A Theoretcal and Emprcal Investgaton, SCIENZE REGIONALI, FrancoAngel Edtore, vol. (1), pages 3-8. Barro, R. J. and Sala--Martn X. (199): Economc Growth, McGraw-Hll, New York. Canton, E., Mnne, B., Neuwenhus, A., Smd, B., and Steeg, M. (): Human Captal, R&D and Competton n Macroeconomc Analyss, ENEPRI, Workng Paper, no. 38, August. Mankw, N. G., Romer D. and Wel D. N. (1992): A Contrbuton to the Emprcs of Economc Growth, Quaterly Journal of Economcs, 17, p Nga R. ş Pssardes C. (4): Structural Change n a Mult-Sector Model of Growth, London School of Economcs, Centre for Economc Performance, CEPR and IZA. Rostow W. W. (192): The Process of Economc Growth, The Morton Lbrary. 11

12 Appendx ya%, ys%, ypps ya%, ys%, ypps 1 ya%, y%, ypps ya%, y%, ypps 1 12

13 Appendx 2 y a% y a_e 1 y a_l y a_u 1 3 y % y _E y _L y _U y s% 1 y s_e y s_l 8 y s_u