Center for Economic Institutions Working Paper Series

Transcription

1 Center for Economc Insttutons Workng Paper Seres CEI Workng Paper Seres, No "The Performance of oregn rms and the Macroeconomc Impact of DI" Kyoj ukao Center for Economc Insttutons Workng Paper Seres Insttute of Economc Research Htotsubash Unversty 2- Naka, Kuntach, Tokyo, JAPAN Tel: ax: e-mal:

2 The Performance of oregn rms and the Macroeconomc Impact of DI Kyoj ukao* Htotsubash Unversty May 2007 Abstract In ths paper, I examne the macroeconomc mpact of nward DI n Japan. rom a general equlbrum pont of vew of the macroeconomy, probably the most mportant host country beneft of nward DI s mprovements n productvty caused by the nflow of manageral resources. In the frst part of ths paper, whch s largely based on the results of ukao, Ito and Kwon (2005), I revew the evdence suggestng that nward DI rases the average total factor productvty of frms n Japan. In the second part, usng a general equlbrum model of an open macroeconomy, I smulate the macroeconomc mpact of an ncrease n the nward DI stock. The results suggest that f Prme Mnster Abe s goal on nward DI, whch s to ncrease the nward DI stock to 5 percent of GDP by the end of 200 s acheved, ths wll help to rase Japan s GDP by percent and real wage rates by 0.56 percent. Dvdend payments abroad by foregn-owned frms and the fall n Japan s foregn nvestment ncome caused by the nflow of captal (or the declne n captal outflows), wll make the ncrease n Japan s GNP (whch ncludes net foregn nvestment ncome) smaller than the ncrease n GDP. The ncrease n GNP wll be 0.25 percent of GDP. *Correspondence: Kyoj ukao, Insttute of Economc Research, Htotsubash Unversty, Naka 2-, Kuntach, Tokyo 86 JAPAN. Tel.: , ax: , e-mal: k.fukao@srv.cc.ht-u.ac.jp

3 . Introducton Accordng to economc theory, foregn drect nvestment (DI) s a form of long-term nternatonal captal movement whch s accompaned by nvestors ntangble assets, such as the accumulated technologcal knowledge through R&D, management sklls, or marketng know-how based on past advertsng actvty. Because of these ntangble assets, afflates owned by foregn frms are expected to enjoy hgher total factor productvty (TP) and proft rates than the average domestc frm. Such nflows of ntangble assets wll beneft the Japanese economy. In ths paper, I quanttatvely evaluate how much the Japanese economy wll beneft from ncreases n nward DI. In the frst half of ths paper, I revew some econometrc evdence from ukao, Ito and Kwon (2005) whch suggests that foregn frms dsplay hgher productvty than domestc frms. The method of nvestgaton n that study s based on the followng reasonng: f foregn frms n Japan possess technologes that are superor to those of ther domestcally-owned counterparts, then ths should manfest tself n hgher total factor productvty (TP). In ths case, Japan benefts from nward DI. Lke DI n other developed economes, the largest part of recent nflows to Japan took the form of mergers and acqustons (M&As). There s, of course, the possblty that foregn frms may enjoy greater productvty because they pck frms wth hgher TP as M&A targets. In order to take account of ths possblty, the study also tests whether foregners have tended to acqure frms that already enjoy hgher TP, or whether the acqured frms productvty mproved after the take-over. The second part of ths paper then uses the results of ukao, Ito and Kown (2005) on the effects of DI on frm-level productvty to estmate the mpact of nward foregn nvestment on Japan s macroeconomy. Snce such nvestments represent nternatonal captal flows accompaned by ntangble assets, t s possble to use standard economc theory on the nternatonal movement of producton factors for the macroeconomc analyss. In the case of DI, the home country wll beneft from the earnngs of ther afflates abroad, whle n the host country, the hgher productvty of foregn frms wll ncrease real wage rates and beneft workers. oregn frms wll have a hgher rate of return to captal because of ther hgher productvty, and ths hgher rate of return to captal wll nduce captal deepenng through captal mports. Ths captal deepenng wll further rase real wage rates and beneft workers n the host country. Usng ths reasonng and a smple standard macro model, the effects of nward DI on Japan s macroeconomy and the balance of payments structure wll be calculated. Several recent studes conductng macroeconomc growth accountng exercses have tred to solate ntangble assets from TP growth and treat them as a factor nput (see, e.g., Corrado et al. 2005, 2006). Snce t s very dffcult to measure the accumulaton of some types of ntangble assets at the frm level, n ths paper, I take the more tradtonal approach of treatng ntangble assets not as factor nputs, so that TP growth ncludes the contrbuton of ntangble asset accumulaton.

4 The analyss n ths paper focuses on the drect benefts of the hgher productvty of foregn frms, whch s relatvely easy to measure. However, n addton, DI may also mprove Japan s socal welfare through a varety other channels. When foregn frms ntroduce new goods and servces, consumers n the host country beneft from these nnovatons In the case of servces, many of whch are non-tradable, customers would not be able to enjoy the new servces offered by foregn frms f these do not set up a supply-base n the country. Inward DI may also enhance competton n the host country market and mprove market performance. nally, Japanese frms may also beneft from technologcal spllovers from advanced foregn frms. These are all pertnent ssues wth regard to nward DI and I hope to address these dffcult-to-measure benefts n the future. 2. Does Inward DI Increase Japan s TP? Concernng the potental benefts of nward foregn drect nvestment for the host economy, t has been frequently clamed that such nvestment, and especally greenfeld nvestment, generates new employment. It has also been argued that export-orented foregn frms help to mprove the host country s current account balance. However, t s mportant to note that these arguments are based on a partal equlbrum approach and do not take account of second-order effects throughout the economy. or example, the ncreased competton from foregn frms may also lead to a reducton of employment at domestcally-owned frms. Moreover, exports by foregn frms may lead to an apprecaton of the host country s currency and therefore potentally do not help to mprove ts current account balance. In the end, standard macroeconomc theory suggests, the current account balance of an economy wth free nternatonal captal flows s determned by the savng-nvestment balance. rom such a general equlbrum pont of vew, probably the most mportant host country beneft from nward DI s the mprovement n productvty brought about by the nflow of manageral resources. In ths subsecton, relyng on the results of ukao, Ito and Kwon (2005; hereafter referred to as ukao et al.), I revew the evdence that foregn frms ndeed show a better performance than the average domestc frm. 2 2 Qute a number of studes, on varous countres, have dealt wth ths topc. These typcally show that labor tends to be more productve n foregn-afflated companes than n domestc companes. See, for example, Blomström and Sjöholm (998) on Indonesa and Grffth and Smpson (200) on Brtan. Doms and Jensen (998) n ther study on the US found that US multnatonal plants had the hghest labor productvty, followed by foregn-owned establshments, whle US-owned non-multnatonal plants had the lowest labor productvty. However, ths s generally due to a greater concentraton of captal nvestment; total factor productvty (TP) analyses ndcate that foregn frms productvty s not necessarly hgher f dfferences n captal ntensty are taken nto account. Studes comng to ths concluson nclude Ito (2004b) on Indonesa, Ramstetter (200, 2002) and Ito (2002, 2004a) on Thaland, and Globerman, Res, and Vertnsky (994) on Canada. 2

5 Comparng the Performance of oregn and Domestc rms n Japan ukao et al. used the frm-level panel data underlyng the Basc Survey of Japanese Busness Structure and Actvtes conducted annually by the Mnstry of Economy, Trade and Industry (METI). The survey covers all frms wth at least 50 employees and 30 mllon of pad-n captal n the Japanese manufacturng and mnng sectors and several servce ndustres. ukao et al. used the data for manufacturng frms. Ther data cover the perod ( n the case of the analyss on M&A). After some screenng of the data ther panel data conssts of 93,880 observatons. 3 In the survey, frms were asked what percentage of ther pad-n captal was owned by foregners and whether they had a foregn parent ownng more than ffty percent of the frm. Based on ths nformaton, ukao et al. determned whether a frm s foregn-owned. They used the followng two defntons of foregn-owned frms: a broad defnton, where one or several foregners own 33.4 percent or more of the frm s pad-n captal n total, and a narrow defnton, where foregn-owned frms are those majorty-owned by a sngle foregn frm. It should be noted, though, that there are several Japanese frms where more than one thrd of ssued stocks are owned by foregn nsttutonal nvestors as portfolo nvestment and there s therefore a rsk that the broad defnton ncludes such frms. In ths paper, I manly refer to the results for the narrow defnton. Table, whch s based on the METI data, shows how the presence of foregn-owned frms n Japan s manufacturng sector ncreased n The number of foregn-owned frms grew from 95 n 994 to 236 n Durng the same perod, the sales of foregn-owned frms nearly doubled from 2.2 trllon to 23.7 trllon. 62 foregn-owned frms exted and 73 foregn-owned frms newly entered n ths perod. 6 domestcally-owned frms n 994 had become foregn-owned by I regard these frms as havng been acqured by foregn frms. INSERT Table The ncrease n foregn-owned frms market share was manly caused by these 6 M&As. The total sales of these 6 frms amounted to 4. trllon n 2000, whch s greater than the total ncrease n 3 Kmura and Kyota (2007), who used the same data source, examned the relatonshp between ownershp and frms performance ndcators (such as the captal-labor rato, real value-added and TP). Coverng the perod (fscal years), ther study shows that foregn-ownershp has a postve mpact on the growth rate of real value-added, the rate of return to captal, and TP. Compared wth Kmura and Kyota s analyss, the ukao et al. study s more sharply focused on the TP level as a measure of performance and the effect of out-n and n-n M&As. 3

6 foregn-owned frms sales of.5 trllon n the perod. ukao et al. measure each frm s TP level usng the method developed by Good, Nadr, and Sckles (997). gure compares the hstograms of foregn-owned and domestcally-owned frms TP. The fgure shows that foregn-owned frms tend to have substantally hgher TP levels than domestcally-owned frms. The dstrbutons are based on pooled data and determnants of the TP level other than foregn ownershp are not taken nto account. Therefore, the derved nterpretaton carres the rsk of beng based. or example, suppose that the average TP level grows over tme and the market presence of foregn-owned frms s also on the rse. In that case, snce observatons for foregn frms are concentrated n the latter part of the observaton perod when the average TP level s hgher, n pooled data for the entre perod, foregn frms wll dsplay hgher average TP than domestc frms even when there s no gap n TP n any partcular year. In order to avod ths knd of bas, ukao et al. conducted a regresson analyss. INSERT gure rms performance s regressed on the foregn-ownershp dummy and frms other characterstcs. As a frst step, only the ndustry and year dummes are used. The man results (usng the narrow defnton of foregn-owned frms) are as follows (Table 2): ) oregn-owned frms TP s about 8 percent hgher and ther current proft-to-sales rato.5 percentage ponts hgher. 2) oregn-owned frms enjoy slghtly hgher TP growth. 3) oregn-owned frms spend proportonately more on R&D per worker. They also have a sgnfcantly hgher captal-labor rato. Probably because of ths, the labor productvty of foregn-owned frms s hgher than that of domestcally-owned frms. 4) There s no sgnfcant dfference between domestcally-owned and foregn-owned frms n the growth rates of real sales and employment. But foregn-owned frms show a sgnfcantly lower growth rate of tangble assets. 5) Average wages at foregn frms are.28 mllon hgher per year. INSERT Table 2 As we have seen, foregn-owned frms tend to conduct more R&D and pay hgher wage rates. Although ther TP level s sgnfcantly hgher than that of Japanese frms, ths dfference mght be 4

7 caused not by the transfer of manageral resources, knowledge, etc., from ther parent frms but by ther own R&D actvtes or the (potentally) hgher qualty of ther labor. In order to test whch of the above two hypotheses s correct, ukao et al. emprcally examne the determnants of each frm s TP level and TP growth rate. Descrptve statstcs of the man varables used n the regresson are presented n Table 3, whle the results of ths regresson are reported n Table 4. The regresson s conducted usng OLS and pooled data for INSERT Tables 3 and 4 Agan, foregn-owned frms dsplay a TP level about 5 percent hgher than that of Japanese frms even after controllng for other factors such as R&D ntensty, the percentage of non-producton workers, the number of years snce the frm was establshed, and frm sze (sales) n addton to ndustry dfferences (ndustry dummes) and the observaton year (Table 4.a). However, when ukao at al. add frm dummes to the regresson model, the gap between the TP level of foregn-owned frms and Japanese frms becomes nsgnfcant. Ths result suggests that the strong correlaton between foregn ownershp and the TP level s at least partly the result of the ntally hgher TP level of the frms later acqured by foregn frms. I wll dscuss ths ssue n more detal later. Table 4.b shows that foregn-owned frms have a.4.8 percentage-pont hgher (annual) TP growth rate than Japanese frms even after controllng for other factors. Yet, ths postve correlaton between foregn ownershp and the TP growth rate agan becomes nsgnfcant n the fxed effect models. Overall, the comparson between foregn-owned and domestcally-owned frms shows that foregnowned companes had a 5 percent hgher TP level as well as hgher returns on captal. Moreover, they dsplayed a hgher captal-labor rato and R&D nvestment per worker. They also enjoyed a hgher TP growth rate. Probably reflectng the hgher levels of captal ntensty and technology, foregn-owned companes showed hgher labor productvty and wage rates as well. But n the fxed effect models, ukao et al. could not fnd a sgnfcant postve correlaton between foregn ownershp and the TP level or growth rate. Are Good rms Chosen as M&A Targets? As ponted out above, there are two possble theoretcal explanatons for the postve correlaton between foregn ownershp and productvty. One potental explanaton s that foregn-owned frms enjoy greater productvty because they choose domestc frms wth hgher TP as M&A targets. I 5

8 call ths mechansm the selecton effect. The alternatve explanaton s that Japanese frms that were acqured by foregn frms receve new technologes and management sklls from ther foregn owners and ths transfer of ntangble assets boosts ther TP. or short, I call ths mechansm the technology-transfer effect. In order to determne whch one of the two effects s responsble for the postve correlaton between foregn ownershp and productvty, ukao et al. conduct two emprcal tests. rst, they estmate a Probt model explanng whether a frm s chosen as an M&A target based on ts TP level and other characterstcs. Second, they test whether the TP of Japanese frms that were acqured by foregn frms mproves after the nvestment. ollowng the narrow defnton of foregn ownershp above, ukao et al. defne out-n M&As as cases where a frm that dd not have a foregn parent frm wth majorty ownershp at tme t comes to have a foregn parent frm wth majorty ownershp at tme t. In order to compare out-n M&As wth n-n M&As (M&As nvolvng only domestc frms), they defne n-n M&As as cases where a frm that dd not have a parent frm wth majorty ownershp at tme t comes to have a domestc parent frm wth majorty ownershp at tme t. Table 5 shows the number of out-n and n-n M&A cases n ukao et al. s dataset. It ncludes 67 cases of narrowly defned out-n M&As and,362 cases of n-n M&As. INSERT Table 5 Usng the panel data of manufacturng frms for , ukao et al. estmated a Probt model explanng whether a frm s chosen as an M&A target based on ts TP level and other characterstcs. The dependent varables are the out-n M&A dummy and the n-n M&A dummy. Each M&A dummy varable takes value one when ths type of M&A occurs. As explanatory varables, ukao et al. use the logarthm of the TP level, the growth rate of TP, frm sze (the number of workers), the current proft/sales rato, the total labltes/total assets rato, year dummes, and ndustry dummes. All the explanatory varables are values at the perod (tme t ) precedng the M&A transacton (tme t). Table 6 shows the estmaton results. The determnants of M&As are surprsngly dfferent for out-n and n-n M&As. In the case of out-n M&As, frms wth hgher TP, a hgher proft rate, and of a larger sze are chosen as targets. In the case of n-n M&As, frms wth a lower proft rate, larger labltes, and of a smaller sze are chosen as targets. In both cases, the growth rate of frms TP (from t-2 to t-) does not have any sgnfcant effect on the selecton. INSERT Table 6 6

9 These results mply that foregn frms acqure Japanese frms that already at the tme of acquston show a better performance. It thus seems that at least part of the hgher TP of foregn-owned frms s the result of the selecton effect. In contrast, n-n M&As tend to dsplay characterstcs of rescue mssons. One possble explanaton s that n-n M&As n Japan are manly conducted wthn vertcal and horzontal keretsu networks and fnancally dstressed small frms are salvaged by other member frms through M&As. Do M&As Improve the Performance of Target rms? Next, I wll explan the technology-transfer effect. ukao et al. examned ths effect by estmatng how the performance of out-n and n-n M&A target frms changes after the acquston. The followng econometrc model s estmated: Y f, t+ τ Y f, t = α + outnm & ADummy f, t + 2nnM & ADummy f, t + γ X, + δ f t j IndustryDummy j, f, t θtyeardummy, + T, f, t + j T where Y f, t denotes the performance of frm f n year t. As Y f, t they used the logarthm of the TP level, the logarthm of the number of workers, and the current proft/sales rato. It s qute lkely that t takes several years for technology-transfer effects to manfest themselves and n order to take account of ths tme lag, the effects two years (τ=) and three years (τ=2) after the acquston are examned. As explanatory varables, they used out-n and n-n M&A dummes n year t, the values of the three performance varables (the logarthm of the TP level, the logarthm of the number of workers, and the current proft/sales rato) n year t-, the R&D/sales rato, the total labltes/total assets rato, ndustry dummes, and year dummes. In the case of the estmaton where changes n employment are the dependent varable, they used sales per worker as an addtonal explanatory varable n order to take account of labor hoardng. ε f, t The regresson results on the effects two years (τ=) after the acquston are reported n Table 7, whle the results on the effects three years later (τ=2) are reported n Table 8. INSERT Tables 7 and 8 The results ndcate that out-n M&As mprove target frms TP level and current proft/sales rato. Compared wth out-n M&As, n-n M&As brng a smaller and slower mprovement n target frms TP level and there s no mprovement n the current proft/sales rato. The mpact of out-n M&As on target frms employment s also sharply dfferent from that of n-n M&As. In the case of n-n M&As, there s a sgnfcant and postve effect on employment two years after the acquston, whle n the case of out-n M&As, the effect on employment s negatve but nsgnfcant. 7

10 Overall, ukao et al. found some evdence showng that target frms TP mproved as a result of out-n M&As. Compared wth n-n M&As, out-n M&As brng a larger and qucker mprovement n TP and the proft rate but, at least n the short-run (.e., two years after the acquston) do not ncrease employment at the target frms. The overall comparson between foregn-owned and Japanese companes shows that foregn-owned companes enjoyed 5 percent hgher TP as well as hgher earnngs and returns on captal. They also dsplayed a hgher captal-labor rato and hgher R&D ntensty. Reflectng ther hgher TP and labor-savng producton patterns, foregn-owned companes showed hgher labor productvty and wage rates as well. By estmatng Probt models, t was found that foregn frms acqure Japanese frms wth hgher TP levels and hgher proft rates. In contrast, n-n M&As seem to have the characterstcs of rescue mssons. Small frms wth a hgher total lablty/total asset rato tend to be chosen as targets of n-n M&As. ukao et al. also estmated the dynamc effects of M&As on target frms. The results ndcate that out-n M&As mprove target frms TP level and current proft/sales rato. Compared wth n-n M&As, out-n M&As brng a larger and qucker mprovement n TP and the proft rate but no ncrease n target frms employment two years after the acquston. To sum up the above results, both the selecton effect and the technology-transfer effect appear to play a role n explanng the postve correlaton between foregn ownershp and productvty. The transfer of ntangble assets from foregn frms to M&A takeover targets represents one mportant avenue by whch Japan can beneft from DI, and the evdence presented here shows that such a technology-transfer effect s ndeed operatng. 3. The Macroeconomc Impact of Inward DI In March 2006, then-prme Mnster Junchro Kozum set a new goal on nward DI. 4 The target was to acheve an ncrease of the nward DI stock to 5 percent of GDP by the end of 20. In hs polcy speech to the 65 th sesson of the Det (September 29, 2006), the new Prme Mnster, Shnzo Abe, promsed to am for the early achevement of ths goal by 200. In ths secton, I estmate the macroeconomc mpact nward DI would have f the target was met. 4 The prevous goal announced by Kozum was to double the nward DI stock from 6.6 trllon to 3.2 trllon n the fve year perod from the end of 200 to the end of By the end of 2005, the nward DI stock stood at.9 trllon, very close to the target. But because of several large retreats, the nward DI stock had plunged to 0.6 trllon by the end of June 2006 (prelmnary estmate by the Mnstry of nance and the Bank of Japan). General Motors reduced ts stake n Suzuk by 0.23 trllon n March, whle Vodafone sold ts portable phone busness to Softbank n Aprl n a deal worth.7 trllon. 8

11 Usng Japan s GDP n 2006 ( 50 trllon) and assumng a nomnal GDP growth rate of 2.0 percent, the country s GDP at the end of 200 should be around 552 trllon. 5 percent of ths the target for the nward DI stock amounts to about 27.6 trllon. At the end of June 2006, the nward DI stock was 0.6 trllon. Therefore, I estmate the macroeconomc mpact of an ncrease n the nward DI stock by 7.0 trllon. However, rather than basng the estmaton of the macroeconomc mpact on the ncrease n the nward DI stock, t s more approprate to consder the ncrease n the presence of foregn-owned frms n the economy ths translates nto. As a frst step, let us estmate how much the employment and the value added of foregn frms would ncrease. Based on mcro-data from the Establshment and Enterprse Surveys for 996 and 200, the ukao et al. study suggests that between these two years, the number of workers employed by the Japanese afflates of foregn frms (33.4 percent or more foregn-owned) ncreased by 27,000, from 485,000 to 756,000. Durng the same perod, the nward DI stock ncreased by 3.6 trllon, from 3.47 trllon to Ths means that an ncrease of the nward DI stock of.7 mllon was needed for each addtonal person employed by a foregn-owned frm n Japan. If we assume that ths relatonshp remaned unchanged from 200 to 2006, then the number of employees at foregnowned frms n Japan as of June of 2006 was an estmated,095,000 (=756,000+(0,600,000 6,630,000)/.7). A further ncrease of the nward DI stock by 7.0 trllon from June 2006 to 200 thus would rase the number of employees of foregn-owned frms by an addtonal,453,000. Accordng to ths back of the envelope calculaton, f Abe s goal s accomplshed, the number of workers employed by foregn-owned frms would more than trple by 200 compared wth the number n 200, from 756,000 to 2,548,000. Usng nformaton on the average per-capta gross value added n each ndustry at the 3-dgt ndustry level, ukao et al. estmated that the 756,000 employees of foregn-owned frms n 200 created gross value added of 8. trllon. That s, the average per capta gross value added of foregn-owned frms was 0.7 mllon. As labor productvty mproves, per capta gross value added usually ncreases over tme. If we assume that the average per capta nomnal gross value added of foregn-owned frms ncreases by 2.5 percent annually from 200 to 200, the estmated,095,000 employees n 2006 created 3.3 trllon of gross value added, equvalent to approxmately 2.6 percent of GDP. 5 And n 200, the estmated 2,548,000 employees of foregn-owned frms would create 34.0 trllon of gross value added, whch would be about 6.2 percent of Japan s projected GDP n that year. As I explaned above, these results do not mean that the addtonal DI wll rase Japan s GDP by as 5 The fgure for Japan s GDP n 2006 s based on the second prelmnary estmate publshed on ebruary 5,

12 much as 3.6 percent. A large part of the ncrease n foregn-owned frms producton wll be canceled out by a correspondng declne n domestc frms producton. rom ths general equlbrum pont of vew, the major beneft of the new DI wll be the mprovement of Japan s total factor productvty. The emprcal results reported n the prevous secton suggest that foregn frms average TP level s 8 percent hgher than that of domestc frms. However, t was also shown that foregn frms enjoy hgher TP levels partly because they purchase Japanese frms that already have hgher productvty than the average frm. 6 But the Japanese frms newly acqured by a foregn frm also dsplay a further 2.2-percent ncrease of ther TP level n comparson wth other domestc frms two years after the acquston (see the coeffcent on the out-n M&A dummy n the frst estmaton n Table 7). Because large-scale DI and out-n M&As are such a recent phenomenon n Japan, and the data used n the emprcal nvestgaton reported above only go up to 200, the tme span covered s relatvely short. It s therefore dffcult at ths stage to assess the long-term effects of acqustons by foregn frms on the TP level of acqured Japanese frms. Based on the consderatons above, the smulaton of the macroeconomc mpact of nward DI consders dfferent scenaros for the ncrease n foregn frms gross value added share n the Japanese economy and the effect of out-n acqustons on TP. or the standard scenaro, let us assume that the ncrease n the share of the gross value added by foregn frms n Japan s total GDP by as much as 3.6 percentage ponts s the result of out-n acqustons. Moreover, let us also assume that out-n acqustons ncrease the TP level of purchased Japanese frms by 5 percent n the long run. A 5-percent ncrease n the TP level means that Japanese frms purchased by foregn frms can produce 5 percent more output from the same amount of nput. Ths change n tself wll ncrease Japan s GDP. But from a general equlbrum pont of vew, many other addtonal changes n the Japanese economy are expected. rst, the mprovement n TP rases the demand for labor and ncreases the real wage rate. Second, the mprovement n TP rases rate of return to captal and wll nduce new captal accumulaton. Ths captal accumulaton wll further ncrease Japan s GDP. Thrd, the ncrease n nvestment wll reduce Japan s savng-nvestment balance, the current account surplus and captal outflows. ourth, dvdend payments abroad by foregn-owned frms and the fall n Japan s foregn nvestment ncome, caused by the declne n captal outflows, mean that the ncrease n Japan s GNP, whch ncludes net foregn nvestment ncome, wll be smaller than the ncrease n Japan s GDP. Takng these addtonal changes nto account, I evaluate the overall effects of nward DI on Japan s GDP and GNP by constructng a general equlbrum model of the economy. We should note that the 6 Because the majorty of nward drect nvestments n developed economes such as Japan are conducted as M&As and not as greenfeld nvestments, t s approprate to concentrate on the TP level of acqured domestc frms here. 0

13 assumptons here, a 5 percent ncrease n the TP level and a 3.6 percentage pont ncrease of foregn-owned frms producton share n GDP from 2006 to 200, are relatvely conservatve. rstly, nward DI mght brng larger TP mprovements. The comparson of the TP levels of Japanese and foregn-owned frms n Secton 2 s based on data for the manufacturng sector. But accordng to several precedng studes (Baly and Solow 200; ukao and Myagawa 2007), Japan s labor productvty level s on par wth the US and major European economes n the case of manufacturng ndustres. On the other hand, Japan s labor productvty level n many nonmanufacturng sectors, such as retal and constructon, s only about two-thrds or less of labor productvty levels n the US and major European economes. Therefore, DI from developed economes n Japan s non-manufacturng sectors would probably brng much larger TP gans than suggested by the estmates above for the manufacturng sector. Secondly, there s consderable scope for the share of foregn owned frms producton to ncrease by much more than the 3.6 percentage ponts, f Japan were to abolsh obstacles to nward DI and the presence of foregn-owned frms ncreased to levels smlar to those n other developed economes. In 2002, foregn-owned frms share n manufacturng turnover was only 2.6 percent n Japan, but 20.3 percent n the US, 24.4 percent n Germany, 35.9 percent n rance, and 36. percent n the UK. In servces, foregn owned frms share n turnover n Japan was 0.9 percent compared wth 7.8 percent n the US, 8.7 percent n Germany, 9.5 percent n rance, and 6.8 percent n the UK (OECD 2005). Thus, even f foregnowned frms share n turnover n Japan were to trple or quadruple, t would stll only be half or less of that n other major economes. Based on these consderatons, four scenaros are consdered. The standard scenaro shows the potental macroeconomc mpact based on the assumpton that the TP mprovement and the ncrease n the presence of foregn-owned frms s relatvely moderate n lne wth the estmaton results for TP and the government target. In addton, however, alternatve scenaros are also consdered. The frst alternatve scenaro assumes that the TP mprovement effect n the servce sector, and hence overall, s probably much greater than n the manufacturng sector. The second alternatve assumes a much larger n ncrease n the presence of foregn frms to llustrate the mpact that DI could have even at levels that would stll lag consderably behnd those of other advanced economes. The last scenaro, fnally, combnes the assumptons of a larger TP mprovement and a larger ncrease n DI. Thus, the followng four cases are examned: Case I (standard scenaro): a 5 percent mprovement of the TP level and a 3.6 percentage-pont ncrease n foregn-owned frms share n producton. Case II: a 5 percent mprovement of the TP level and a 3.6 percentage-pont ncrease n foregnowned frms share n producton (.e., t s assumed that the mprovement of the TP level s three tmes as great as n the standard scenaro).

14 Case III: a 5 percent mprovement of the TP level and a 0.8 percentage-pont ncrease n foregnowned frms share n producton (.e., t s assumed that the ncrease n foregn frms presence s three tmes as great as n the standard scenaro). Case IV: a 5 percent mprovement of the TP level and a 0.8 percentage-pont ncrease n foregnowned frms share n producton. The detals of the analyss are reported n Appendx A. The man assumptons on whch the macroeconomc model s based can be summarzed as follows: () In order to explan the coexstence of productve and unproductve frms, t s necessary to assume that frms produce dfferentated products. It s also assumed that frms operate under monopolstc competton. In such an economy, out-n acqustons may brng two types of nnovaton. Usng foregn parent frms technology, domestc frms may mprove ther producton process (process nnovaton). Alternatvely, they may ncrease the varety of ther products (product nnovaton). In order to smplfy the analyss, I assume that only process nnovatons wll occur. (2) The analyss s statc, ths s, t only examnes the long-run effect of a one-tme acquston of domestc frms by foregn frms. In order to smplfy the analyss, the employment and producton share of foregn-owned frms n the Japanese economy are assumed to ncrease from zero to 3.6 percent or 0.8 percent. (3) All products are nternatonally traded wthout trade costs. Very smooth nternatonal ndrect captal flows ensure that Japan s real nterest rate s equal to the world real nterest rate. (4) The world equlbrum nterest rate and the world prce level wll not be affected by the ncrease n Japan s nward DI. The results of the smulaton analyss are shown n Table 9. The second column of the table shows the results for Case I: a 5 percent mprovement of the TP level and a 3.6 percentage pont ncrease n foregn-owned frms share. The second row of the table shows that n Case I, the ncrease n nward DI wll rase Japan s real wage rate by 0.7 percent. As I explaned n Secton, DI conssts of nternatonal captal flows accompaned by the transfer of ntangble assets. Accordng to standard economc theory on the nternatonal movement of producton factors, n the DI host country, the hgher productvty of foregn frms wll ncrease real wage rates and beneft workers. The hgher rate of return to captal wll nduce captal deepenng through captal mports, and ths captal deepenng wll also rase real wage rates and beneft workers n the host country. The 0.7 percent ncrease of the real wage rate shown n Table 9 s the result of all these effects combned. In our macro-model, the rato of the ncrease n the captal stock to the ntal captal stock s also 0.7 2

15 percent. The ncrease of the average level of frms productvty and the captal deepenng nduced by the hgher rate of return to captal together rase Japan s GDP by 0.24 percent n Case I. Ths s a permanent ncrease,.e., wages and GDP wll be hgher by ths amount n every year. When we assume that nward DI rases the TP level three tmes as much (Case II) or the ncrease n foregn-owned frms share s three tmes as great (Case III), the mpact n terms of the ncrease n the wage rate, nduced captal deepenng, and the ncrease n GDP almost trples n comparson wth Case I. That s, the sze of the macroeconomc mpact s almost proportonate to the magntude of ether the TP mprovement or the ncrease n nward DI. In the case of the most optmstc scenaro, Case IV, we can expect large ncrease n real wages and GDP. INSERT Table 9 Dvdend payments abroad by foregn-owned frms and the fall n Japan s foregn nvestment ncome caused by the nflow of captal (or the declne n captal outflows) mean that the ncrease n Japan s GNP (whch ncludes net foregn nvestment ncome) wll be smaller than the ncrease n GDP. How the ncrease n GNP wll dffer from the ncrease n GDP s also shown n Table 9. Inward DI wll change Japan s nternatonal asset-lablty poston n three respects: frst, foregn nvestment ncreases Japan s nternatonal labltes and overseas nvestors wll receve the profts earned by foregn-owned frms n Japan. Under our assumptons, whch are explaned n Appendx A, dvdend payments from foregn-owned frms n Japan to foregn nvestors wll be 0.36 percent of total GDP n Case I. Second, n compensaton for the sale of Japanese frms, Japanese resdents wll receve foregn assets. It s assumed that Japanese resdents use ths money as portfolo nvestment abroad. Investment ncome from Japan's assets abroad receved n compensaton for the sale of Japanese frms wll be 0.30 percent of total GDP. Thrd, the captal accumulaton nduced by nward DI wll cause captal nflows and ncrease Japan s labltes to non-resdents. Interest payments abroad for Japan s labltes resultng from the captal deepenng nduced by nward DI wll be 0.05 percent of total GDP. These three changes n Japan s nternatonal asset-lablty poston wll permanently change Japan s balance of payments. Taken together, these effects wll reduce Japan s ncome account surplus by 0. percent of Japan s GDP n Case I. Thus, subtractng ths ncome account effect from the ncrease n GDP through nward DI, the ncrease n GNP wll be =0.3 percent of GDP. In Case III, whch assumes that the ncrease n foregn-owned frms share s three tmes as large as n the standard scenaro (Case I), the change n the balance of payments and n GNP are also about three tmes as large. On the other hand, n Case II, whch assumes an mprovement of TP through nward DI three tmes as great as n the standard scenaro, the effect on the balance of payments s 3

16 slghtly less than three tmes as large and the ncrease n GNP therefore slghtly more than three tmes as large.. 4. Conclusons In ths paper, I examned the macroeconomc mpact of nward DI n Japan. rom a general equlbrum pont of vew, probably the most mportant host country beneft of nward DI s the mprovement n productvty caused by the nflow of manageral resources. In the frst part of the paper, manly referrng to the results of ukao, Ito and Kwon (2005), I revewed the evdence suggestng that nward DI wll rase the average TP level of frms n Japan. In the second part of the paper, usng a general equlbrum model of an open macroeconomy, I smulated the macroeconomc mpact of an ncrease n the nward DI stock. I found that f Prme Mnster Abe s goal on nward DI, whch s to ncrease the nward DI stock to 5 percent of GDP by the end of 200, s attaned, Japan s GDP wll ncrease by 0.24 percent and Japan s real wage rate wll rse by 0.7 percent. Dvdend payments abroad by foregn-owned frms and the fall n Japan s foregn nvestment ncome caused by the nflow of captal (or the declne n captal outflows), wll make the ncrease n Japan s GNP (whch ncludes net foregn nvestment ncome) smaller than the ncrease n Japan s GDP. The ncrease n GNP wll be 0.3 percent of GDP. 4

17 References Baly, M.N. and R.M. Solow, (200) Internatonal Productvty Comparsons Bult from the rm Level, Journal of Economc Perspectves, vol. 5, pp Blomström, M. and. Sjöholm. (998) Technology Transfer and Spllovers? Does Local Partcpaton wth Multnatonals Matter? NBER Workng Paper, no. 686, Natonal Bureau of Economc Research, Cambrdge, Mass. Corrado, C., C. Hulten and D. Schel (2005) Measurng Captal and Technology: An Extended ramework, n C. Corrado, J. Haltwanger, and D. Schel (eds.), Measurng Captal n the New Economy, Unversty of Chcago Press, Chcago. Corrado, C., C. Hulten and D. Schel (2006) Intangble Captal and Economc Growth, NBER Workng Paper No. 948, Cambrdge, Mass.: Natonal Bureau of Economc Research. Doms, M. E. and J. B. Jensen (998) "Comparng wages, sklls, and productvty between domestcally and foregn-owned manufacturng establshments n the Unted States," n R. E. Baldwn, R. E. Lpsey and J. D. Rchardson (eds.), Geography and Ownershp as Bases for Economc Accountng, Chcago: Unversty of Chcago Press. ukao, K. and T. Myagawa (2007) Productvty n Japan, the US, and the Major EU Economes: Is Japan allng Behnd? Paper presented at the Thrd Internatonal Workshop of the Jont Research Study Group EU Economy, EUIJ Tokyo Consortum, Htotsubash Unversty, Tokyo, March 25, ukao, K., K. Ito and H.U. Kwon (2005) Do Out-n M&As Brng Hgher TP to Japan? An Emprcal Analyss Based on Mcro-data on Japanese Manufacturng rms, Journal of the Japanese and Internatonal Economes, vol. 9, pp Globerman, S., J.C. Res and I. Vertnsky (994) The Economc Performance of oregn Afflates n Canada, Canadan Journal of Economcs, vol. 27, pp Good, Davd H., M. Ishaq Nadr, and Robn C. Sckles (997), Index Number and actor Demand Approaches to the Estmaton of Productvty, Handbook of Appled Econometrcs Vol. 2, Mcroeconomcs, pp Grffth, R. and H. Smpson (200) Characterstcs of oregn-owned rms n Brtsh Manufacturng, IS Workng Papers 0/0, The Insttute for scal Studes, London. Ito, K. (2002) Are oregn Multnatonals More Effcent? Plant Productvty n the Tha Automoble Industry, ICSEAD Workng Paper Seres, no , The Internatonal Centre for the Study of East Asan Development, Ktakyushu. Ito, K. (2004a) oregn Ownershp and Plant Productvty n the Tha Automoble Industry n 996 and 998: A Condtonal Quantle Analyss, Journal of Asan Economcs, vol. 5, pp Ito, K. (2004b) oregn Ownershp and Productvty n the Indonesan Automoble Industry: Evdence from Establshment Data for , n T. Ito and A. Rose (eds.), Growth and 5

18 Productvty n East Asa, Unversty of Chcago Press, Chcago and London. Kmura,. and K. Kyota (2007) oregn-owned versus Domestcally-Owned rms: Economc Performance n Japan, Revew of Development Economcs (), OECD (2005) Measurng Globalsaton: OECD Globalsaton Indcators, OECD, Pars. Ramstetter, E. D. (200) Labor Productvty n oregn Multnatonals and Local Plants n Tha Manufacturng, 996 and 998, ICSEAD Workng Paper Seres 200-3, The Internatonal Centre for the Study of East Asan Development, Ktakyushu. Ramstetter, E. D. (2002) Does Technology Dffer n Local Plants and oregn Multnatonals n Tha Manufacturng? Evdence from Translog Producton unctons for 996 and 998, ICSEAD Workng Paper Seres , The Internatonal Centre for the Study of East Asan Development, Ktakyushu. 6

19 Appendx A. Macroeconomc Smulaton Analyss of the Impact of Inward DI n Japan Ths appendx provdes the detals of the macroeconomc smulaton of the mpact of nward DI based on a model of the Japanese economy wth mcroeconomc foundatons. Basc assumptons on market and technology The smulaton s based on a model wth monopolstc competton. It s assumed that n Japan, n commodtes are produced by n frms and each commodty s produced by one frm. There are nθ foregn frms and n( θ ) domestc frms. All products are fnal goods and nternatonally traded and there are no trade costs. All consumers, domestc and foregn, have dentcal homothetc preferences wth regard to these goods. Each frm faces the followng demand functon: p σ X = E () p * where X denotes the demand for frm s product, p, s the prce of frm s product, and p* stands for the world prce level. /( σ) s the prce demand elastcty. It s assumed that /( σ) s greater than one. The parameter E denotes the sze of world wde demand. It s also assumed that Japan s not a large country and p* and E can be treated as constant over tme. A Cobb-Douglas constant return producton functon s assumed: K X = a L (2) where L and K denote frm s labor and captal nput and a denotes the TP level of frm. Labor and captal markets are compettve so that the cost share of labor s equal to the constant parameter. The ntroducton of manageral resources through out-n acqustons can be expressed n the model n at least two ways. rst, foregn frms can produce wth a hgher TP level, a (process nnovaton). Second, foregn frms can produce new goods, so that the number of commodtes produced n Japan, n, ncreases (product nnovaton). or smplcty, the analyss here focuses on the frst type of nnovaton as a result of manageral resource transfers. In the baselne scenaro, t s assumed that foregn frms TP level, a, s 5 percent hgher than domestc frms TP level, a J. Let r denote the constant world equlbrum real nterest rate. urther assumptons are that domestc net savng s zero under r and that there s no captal deprecaton. 7

20 The proft maxmzaton behavor of frms In producton functon (2), margnal cost does not depend on the producton level and can be expressed by wl + rk B X = a w r (3) where w denotes the domestc real wage rate and B stands for (/( )) + (( )/). When frms operate under monopolstc competton and the prce demand elastcty s /( σ), proft maxmzng frms set the prce level equal to margnal costs tmes /σ. These condtons yeld the followng optmal output and factor nput levels: X B = w aσ p * σ r E (4) L = a r w X (5) K = a w r X (6) oregn frms, whch have a hgher TP level, a, and can produce output at lower margnal cost, set lower sales prces and produce more output than domestc frms (equaton (4)). The total sales of foregn frms are (a /a J ) σ( σ) tmes greater than the total sales of domestc frms. And both foregn frms share n total sales and ther share n labor nput n the Japanese economy overall are equal to θ (a /a J ) σ( σ) /( θ +θ (a /a J ) σ( σ) ). The labor market equlbrum condton The equlbrum condton for the labor market can be expressed by ( θ ) nl + θ nl L. J = Usng equatons (4), (5), and the above equaton yelds 8

21 9 ( ) L r w p a B w r a n r w p a B w r a n J J = + σ σ σ θ σ θ * * (7) where L denotes Japan s total labor endowment, whch s assumed to be constant. The domestc real wage rate s determned by the above equaton. Equaton (7) shows that nward DI (an ncrease n θ ) wll ncrease the demand for labor and rase the equlbrum real wage rate. Snce the prce level p* remans unchanged, the level of workers welfare wll be mproved by nward DI. Captal accumulaton nduced by DI Productve foregn frms enjoy hgher returns to captal and ths fact wll nduce captal accumulaton. Under producton functon (2), the followng equaton holds: = rk wl (8) where L and K denote Japan s total endowment of labor and total nput of captal. Under the assumpton of perfect nternatonal captal moblty, the captal nput level, K, s endogenously determned by the above equaton. Therefore, the ncrease n captal nput s proportonal to the ncrease n the real wage rate. Japan s GDP before and after nward DI Usng the model just developed t s possble to determne how nward DI changes Japan s GDP. To smplfy the analyss, an equlbrum wthout nward DI (θ =0) and another equlbrum wth nward DI (θ >0) are compared. When there s no nward DI, all frms have the same productvty level and produce the same amount. Therefore, the real GDP level can be expressed by

22 20 = = = 0 0 K L a n K n L na X J J n (9) where K 0 denotes the total nput of captal stock n Japan. When DI occurs and θ n frms are purchased by foregn frms, the real GDP level can be expressed by ( ) ( ) ( ) ( ) θ θ = + = ) ( J J J n K L na K L na X (0) where foregn and domestc frms factor nputs can be expressed by + = for n L a a a a L J J σ σ σ σ θ θ () D + = for n L a a L J σ σ θ θ (2) + = for n K w w a a a a K J J 0 0 σ σ σ σ θ θ (3) D + = for n K w w a a K J 0 0 σ σ θ θ (4) where and D stand for the set of foregn frms and the set of domestc frms. w 0 and w denote the wage rate before and after the nward DI, whch can be calculated usng Equaton (7). Assumptons on parameter values By settng the parameter values, t s possble to smulate the macroeconomc mpact of nward DI.

23 The cost share of labor,, s assumed to be equal to 2/3 and the prce demand elastcty, /( σ), to be equal to 5 (ths means that the mark-up rate wll be (/σ-)*00=25 percent). As explaned n Secton 4, four cases are examned. Here, the calbraton for Case I s explaned n detal. In Case I, t s assumed that foregn frms TP s 5 percent hgher than domestc frms TP,.e., a /a J =.05. The rato of foregn frms to total frms after the nward DI, θ, s set at a level whch satsfes the followng condton: θ (a /a J ) σ/( σ) /( θ +θ (a /a J ) σ/( σ) )=0.36 The above equaton means that after the nward DI took place, foregn frms share of total sales and of labor nput n the Japanese economy overall wll be equal to 3.6 percent. The soluton of the above equaton s θ = Smulaton results The results of the smulaton analyss are shown n Table 9. They show that the ncrease n nward DI assumed n Case I wll rase the real wage rate and total captal nput by 0.7 percent and lft GDP by 0.24 percent. Wages and GDP wll be permanently hgher by ths amount. The effects on Japan s balance of payments Dvdend payments abroad by domestc frms and the fall n Japan s foregn nvestment ncome caused by the nflow of captal (or the declne n captal outflows) mean that the ncrease n Japan s GNP (whch ncludes net foregn nvestment ncome) wll be smaller than the ncrease n Japan s GDP. Table 9 shows how the ncrease n GNP wll dffer from the ncrease n GDP. Inward DI wll change Japan s nternatonal asset-lablty poston n three respects. rst, nward foregn nvestment wll ncrease Japan s nternatonal labltes, snce foregn nvestors wll receve part of the profts earned by foregn owned frms n Japan. Typcally, foregn nvestors receve only part of the profts because when the foregn captal partcpaton rate n a partcular frm s less than one, then domestc nvestors wll also receve ther share. In addton, the Japanese government wll receve corporate ncome taxes from frms n Japan. Second, n compensaton for the sale of Japanese frms, Japanese resdents wll receve foregn assets. It s assumed that Japanese resdents use these funds for portfolo nvestment abroad. Thrd, captal accumulaton nduced by nward DI wll cause captal nflows and ncrease Japan s labltes to non-resdents. These changes n Japan s nternatonal asset-lablty poston wll permanently change Japan s balance of payments. In the smulaton, the effects of these changes are calculated as follows: 2

24 () Dvdend payments to foregn owners In the model, 20 percent of GDP s monopolstc rent. To smplfyng the analyss, t s assumed that frms pay the entre rent to ther stockholders as dvdends and all the real captal accumulaton s fnanced through ndrect fnancng. If the condtons for the Modglan-Mller theorem hold, then frms fnancal structure wll not change the results. It s assumed that the average captal partcpaton rate of foregn parent frms n ther Japanese afflates s 50 percent. It s also assumed that there s not corporate ncome tax. In Case I, after the DI, foregn frms market share s 3.6 percent. Therefore, foregn parent frms receve GDP n annual dvdends, where GDP denotes Japan s GDP after the nward DI. Therefore, gven the smulated ncrease n GDP as a result of nward DI by 0.24 percent and usng the relatonshp GDP /GDP 0 =.0024, dvdend payments to foregn owners n terms of GDP 0 are GDP = GDP 0. (2) oregn ncome from Japan s portfolo assets abroad receved n compensaton for the sale of Japanese frms Before the nward DI, the total value of Japanese frms stocks s 0.2 GDP 0 /r, where r denotes the world equlbrum nterest rate. 50 percent ownershp of θ percent of Japanese frms s sold to foregn frms. In exchange, Japanese resdents receve 0.5 θ 0.2 GDP 0 /r n foregn assets. The annual nvestment ncome from these assets wll be 0.5 θ 0.2 GDP 0 =0.003 GDP 0. (3) Interest payment abroad for Japan s labltes created by the captal accumulaton nduced by nward DI Addng the smulated ncrease n Japan s captal stock through DI of 0.7 percent to the exstng captal stock (+0.007), the share of the foregn captal stock as a result of ths DI s 0.7 percent (.e., 0.007/(+0.007)=0.7 percent). Interest payments abroad for these labltes on Japan s part are r K. Snce the cost share of captal s and the mark-up rate s 25 percent, the nterest payments are equal to GDP = GDP 0. Taken together, these effects wll reduce Japan s ncome account surplus by =0. percent of Japan s GDP. Therefore the ncrease n GNP wll be =0.3 percent of GDP. 22

25 Source: ukao, Ito and Kwon (2005). 23