Imports and Growth in China1 M.J. Herrerias† and V. Orts‡ †Department of Economics, Universitat Jaume I ‡Department of Economics and Institute of International Economics, Universitat Jaume I 31 July 2009 Abstract Capital accumulation and export promotion policy have been considered in the literature one of the main sources of rapid economic growth in China. However, while endogenous growth models emphasize the role of imports, rather than exports, in economic growth, to the best of our knowledge there is no empirical evidence that analyses the importance of imports as a source of long-run growth to boost productivity and economic development in China. This study attempts to cover that gap. Thus, it aims to explore whether imports and investment could be determinants of output and labour productivity in the short and long run in China for the period 1964-2004. Furthermore, given that some authors have argued that trade is more a consequence of the process of growth than a cause, this paper also seeks to analyse the interaction and the long-run causality among imports, investment, output and productivity. Therefore, we examined whether the rapid growth process was mainly explained by an importled growth or, on the contrary, economic growth and trade in China are driven just by investment. The empirical results provide evidence that both imports and investment encourage output and labour productivity in the long run, but neither investment causes imports nor imports cause investment. Keywords: Imports, Investment, R&D, Growth, China JEL classification: F43, O40, O47, O53

1

Contact: Department of Economics, Universitat Jaume I, Avda. Sos Baynat s/n, 12071 Castellón (Spain). Tel.: +34 964 38 71 77; fax: +34 964 72 85 91. Email: [email protected] . [email protected] The authors gratefully acknowledge financial support from the Spanish Ministry of Science and Innovation, (project ECO2008-06057/ECON) and the Generalitat Valenciana. The usual disclaimer applies.

1. Introduction

In economic terms, the growth of China has been remarkable for almost four decades. The literature has considered capital accumulation and export promotion policy to be the main sources of this rapid economic growth (Chow, 1993; Siebert, 2007). The Chinese economy, with its singular characteristics, followed the strategy begun by other rapidly developing Asian countries (East Asian Miracle countries) that highlight their rapid export promotion as a central channel enhancing economic growth (World Bank, 1993). However, the endogenous growth literature, in line with the models proposed by Grossman and Helpman (1991), Lee (1995) and Mazumdar (2001), to cite just a few, emphasizes the role played by imports rather than exports in economic growth. In these models, imports (through access to capital goods and intermediate goods from technologically more advanced countries) have become a form of technology transfer and a source of competition that stimulates the domestic industry. Nevertheless, there are other studies, like Rodrik (1995), which suggest that the increase in growth rate in Asian countries was mainly in response to variations in investment, trade being a consequence rather than a cause of rapid economic growth. To the best of our knowledge there is no empirical evidence that analyses the importance of imports as a source of long-run growth in China. This study attempts to cover this gap in the empirical literature on China’s economic growth; that is, its aim is to analyse the role played by imports in boosting productivity and economic development in this country. China is an interesting case of study because, in spite of the general perception about the decisive role played by exports in the process of growth, in our view, this was not the only factor responsible for its fast growth over the last four decades. We think that the importation of foreign technology has played a key role in the process of industrialization in the Chinese economy since the fifties. In fact, one of the main objectives of the Chinese government has been to gain access to advanced foreign technology and equipment. This strategy (with its singularities) predominated throughout the period under study (1964-2004). First, in the mid-sixties, there was a change in the suppliers of imported capital goods, from Russia to Western countries (Japan, USA and European countries), which facilitated access to more advanced technology. Second, during the pre-reform period, the strategy of imports was concentrated on the importation of complete plants2 and equipment to establish the productive capacity, and during the post-reform period it was concentrated on renewing and updating existing obsolete production facilities. In 1980s the Chinese policy on technology imports changed significantly to become “in line with and a part of the overall economic reform programme and the ‘open-door’ policy”.3 Although imports diversified in comparison to the pre-reform period, capital goods, as key pieces of equipment and production lines, still accounted for a very large share of foreign exchange spending (Shi, 1998). These changes attempted to make a more efficient use of economic resources. Finally, the decentralization process and the market forces were gradually introduced to replace the central planning. This period was characterized by an increase in the presence of the non-state sector (Township and Village Enterprises and foreign investment). In this paper we analyse the role of imports and investment in labour productivity and output in China for the period 1964-2004. However, as we know that there could be 2 3

In the 1950s many investment projects imported from Russia were incomplete. Shi (1998), p. 2

2

other factors influencing economic development, we have also considered the role played by domestic innovation activities, competitiveness and foreign economic conditions. Furthermore, we also examined whether the rapid growth of the Chinese economy since the mid-sixties can be explained mainly by an import-led growth effect (in line with the predictions of the endogenous growth theory) or, on the contrary, it is investment that drives economic growth in China, and trade is more a consequence of growth than a cause, as Rodrik (1995) suggested. The empirical results provide evidence that both imports and investment encourage output and labour productivity in the long run. However, neither investment causes imports nor imports cause investment, both playing an independent and positive role as sources of economic growth. Moreover, we also found that R&D expenditure encourages investment in the long run. Thus, this paper contributes to the literature in providing new evidence of the role played by imports in the process of Chinese growth, while the majority of works in the empirical literature have focused on exports. Secondly, in doing so, our analysis focuses on the role of technological progress incorporated into the Chinese economy through capital accumulation and the imports of capital goods and intermediate inputs, which could be a cause of significant technology transfer from abroad that facilitated industrialization and rapid growth in China. Finally, we employed the VAR model to avoid the endogeneity bias in our estimates, given that it is based on a joint modelling of all the variables considered. Furthermore, this methodology can distinguish between the long-run and short-run effects. The rest of the paper is organized as follows. Section 2 contains the literature review. Section 3 shows the description of the variables considered and the methodology. Section 4 presents the empirical results. Comments and conclusions are given in Section 5.

2. Literature Review

Economists have been interested in the differences in growth rates across countries and the causes that lead to some countries growing more than others for some time. To examine these issues, they have employed different theoretical frameworks, from the neo-classical growth model up to more recent models based on endogenous economic growth. Both approaches consider technological progress to be a key factor in enhancing long-run growth, but while technological progress is considered to be exogenous in the traditional model of growth, in the endogenous growth models technological progress is not considered as a purely random process but rather as one that is determined by the internal forces of the system. In particular, the endogenous growth theory grants a greater role to technological progress, in both developed and developing countries. We can find technological progress embodied in capital goods, in formal innovation activities, in the abilities of human capital or in improved efficiency in the organization of production. However, as argued by Grossman and Helpman (1991), in the less developed economies, the scant activity in commercial R&D or the scarcity of original discoveries that are relevant to the world economy could make us believe that technological progress does not play a significant role in the growth and development of an emerging economy. But as these same authors remark: “Yet the process of industrialization in these countries does involve substantial technical change, in the sense that producers gain mastery over

3

products and processes that are new to the local economy”.4 Of course, developing countries are unable to produce most of the machinery and equipment required by the industrialization process, and their economic growth depends on imports of such goods. In this way, imports play a similar role to that of R&D activities in developed countries, that is, they help developing economies to be able to acquire foreign technology from R&D intensive countries (Coe et al., 1997; Lee, 1995; Mazumdar, 2001; Eaton and Kortum, 2001; Caselli and Wilson, 2004). But even if less developed economies import foreign technologies, the process of assimilating and spreading these technologies to the rest of economy, and thus the effect on own growth, depends on a large variety of factors such as initial technological capabilities of the domestic economy, the role played by the government in negotiating with the technology suppliers, the local efforts in promoting the learning process of the foreign technology, the degree of competition and market liberalization, the macroeconomic stability, and so one. Thus, to be successful, import policy and economic development strategies must be linked, and this is the case in the Chinese economy. The majority of China’s imports consist in capital goods (machinery and equipment) and intermediate goods, and its technology import policy, with the changes that were introduced over the period considered (especially in the late 1970s), has been aligned with the domestic economic objectives and is considered to be a key component of the country’s economic development strategy. Thus, the different reform programmes have been aimed at improving the efficiency and assimilation of technology imported from abroad.5 This policy option has a serious shortcoming, however: it is conditioned by the foreign exchange constraint.6 Thus, countries that are involved in this import policy have to develop an aggressive export policy, open the economy to international capital flows, or both. Furthermore, the contribution of imports to industrialization and growth in less developed countries requires a reallocation of resources and an increase in investment. In this effort, imports play an additional role, namely, that of improving the efficiency of capital accumulation by importing relatively cheaper capital goods from high-income countries that are intensive in R&D (Lee, 1995). The same occurs when they have access to an increasing variety of higher quality intermediate inputs in foreign markets (Grossman and Helpman, 1991; Amiti and Konings, 2005). Additionally, there is empirical evidence that suggests that those who participate in international transactions are more likely to survive than those not involved in international trade, given that it forces domestic firms to improve their efficiency (Lopez, 2006). The strong competitiveness and the spillovers generated by technological progress embodied in imports may favour the innovation of new products and processes in the domestic economy (Grossman and Helpman, 1991; Traca, 2002).7 These technological spillovers improve efficiency and consequently enhance growth (Rivera-Batiz and Romer, 1991; MacDonald, 1994; Amiti and Konings, 2005). However, while the arguments discussed above support the idea that the importation of technology (machinery and equipment) and access to intermediate goods enhances economic growth, Rodrik (1995) argues that trade is more a consequence of the process 4

Grossman and Helpman (1991), p. 12. See Conroy (1986). 6 See Mazumdar (2001), Lee (1995) and Wall (1968). 7 There are other externalities associated to the importing activity, like learning-by-doing, improving the managerial effort, organizational structures and foreign contacts, and so on. 5

4

of growth than a cause. In particular, the author has doubts about the effectiveness of trade promotion policies as the main determinant of long-run growth and he argues that the rapid growth experienced by the Asian countries is caused fundamentally by increases in investment that in turn stimulates imports. Moreover, these theoretic discrepancies have not found a suitable answer from the empirical perspective. In a recent study, Krishna, Ozyildirim and Swanson (2003) looked at the incidence of causation and reverse causation (from a short-run perspective) between income, export, import and investment growth in 39 developing countries for the period 1951-1998. The findings of this research yielded mixed results, both as regards the variables that best explain GDP growth (trade variables or investment)8 and with respect to reverse causation.9 They conclude that the direction of causality largely depends on countryspecific characteristics. The evidence is not very clear when we look at the direction of the causality in the long run. On the one hand, evidence supporting the import-led growth effect can be found in Thangavelu and Rajaguru (2004) for India, Indonesia, Malaysia, Philippines, Singapore and Taiwan. Similar findings are to be found in Awokuse (2007) for Poland and in Awokuse (2008) for some South American countries. On the other hand, in Awokuse (2007) the causality is found to run in the opposite direction for the Czech Republic. Finally, and to the best of our knowledge, there is apparently no empirical evidence on the role played by imports on economic growth in China. Thus, analysing the interrelationships and the direction of causality between imports and investment, and between these and economic growth is interesting from both the academic and the policy-makers’ point of view, and can help us to determine the most suitable orientation for economic policies in developing countries. 3. Data and Econometric Issues

We employed annual data for the period 1964-2004 from the National Bureau of Statistics of China (NBS). The period taken for analysis started in the middle of the 60s for two reasons. The first reason was to avoid the turbulent period of the 50s and the economic consequences of the Great Leap Forward that produce abnormal values and make it difficult to perform an empirical analysis.10 The second reason was the change in import policy carried out by the Chinese government at that time. These data are based on the latest compilation published by the NBS in 2004. However and in spite of the continuous efforts to improve the Chinese national accounts, the quality and the accuracy of the statistics have been subject to criticism by many economists.11 The data we are using are more accurate compared to those from previous publications and they have been revised for the whole period following international standards. In addition, it is widely agreed that even though the data present some inaccuracies on certain levels, the long-run trends are approximately correct (Chow, 2004; Holz, 2005 and Bai et al., 2006). Finally, other critics have raised objections about including both the pre-reform and the post-reform period in the same study. In this regard, there are many examples in the literature that use data from the

8

In 19 out of 39 countries, growth is better explained by models which include imports. In 10 out of 39 countries, imports are better explained by models which include GDP growth and 5 of them exhibit two-directional causality. 10 See Chow (1993). 11 See Chow (2006) for this discussion. 9

5

whole period, including Chow (1987), Li (2000) or Yao (2000), among others.12 Nevertheless, we investigate the possibility of structural breaks in our relationships by including the appropriate dummies and testing their significance. Taking the previous considerations into account, our dataset consists of GDP (gdp), labour productivity – Output per worker – (prod),13 imports (im), investment – gross fixed capital formation – (inv), R&D expenditure (rd),14 real exchange rate (rer)15 as a proxy for competitiveness, and US GDP (gdpusa) as a measure of foreign economic conditions. All variables are in both natural logarithmic and real terms.16 The potential interdependence among the relevant determinants of economic growth suggests a joint modelling so as to avoid the endogeneity bias in our estimates. The cointegrated VAR model can solve this problem, as it is based on a simultaneous estimation where the endogeneity or exogeneity of our variables is explicitly examined. Specifically, the methodology we are going to use is based on the cointegrated VAR model proposed by Johansen (1988 and 1995), Johansen and Juselius (1990), and Juselius (2007). We start the analysis with the more parsimonious approach, in which certain restrictions of a statistical and an economic origin will be imposed until the most irreducible form is reached. We consider that this methodology is appropriate, given the aforementioned potential interdependence between the different variables considered. Furthermore, it allows a distinction to be drawn between long-run relationships among the variables under consideration and the determinants of the dynamics of labour productivity and output.17 We can specify the unrestricted VAR model with a linear trend that is restricted to the cointegration space and an unrestricted constant (µ) as follows:18 Y    k −1 k −1 k −1 ~'  Z  ∆Yt = αβ i   + Γi ∆Yt −i + ω∆Z t + ωi ∆Z t −i + θ∆Dst + θ i ∆Dst −i + γDt + µ + ε t t i =1 i =1 i =1   D   s  t −1 ε t ∼ NIID(0, Ω) t = 1....T

∑

∑

∑

(1)

12

Chow (1987) analyses the money supply, prices and income relation for the period 1954-1983. Li (2000) evaluates empirically whether GDP and its sectoral components from 1952 to 1998 can be modelled more accurately as a stationary process around a breaking trend function as opposed to a unit root process. Finally, Yao (2000) analyses how agriculture has contributed to China’s economic development over the period 1952-1999. 13 In this paper labour productivity was corrected by applying the methodology suggested by Nielsen (2004). 14 We took total expenditure on scientific research from NBS as a proxy variable of R&D expenditure. We deflated R&D expenditure with the GDP deflator. 15 The real exchange rate was calculated using the nominal exchange rate between the Chinese currency and the US $ (Renminbi/$) and the respective consumer price indices (CPIs). 16 All the variables were deflated by the GDP deflator. 17 It is possible to find other works using a similar methodology to analyse different aspects of the Chinese economy, for instance Chow (1987), Li (2000), Yao (2000), Phylaktis and Girardin (2005) or Narayan and Sun (2007), among others. 18 The reason for including a trend in the cointegration space is that when the data are following a distinct trend we need to allow for linear trends in the cointegration relationships when testing for the cointegration rank. Otherwise, the estimation would be biased in the absence of a trend in our models.

6

where, αβi are the coefficients of the long-run matrix; α gives the direction and speed of adjustment toward equilibrium and βi are the coefficients of the cointegrated vectors; Yt1 is the matrix of endogenous variables in the model; Zt-1 is the matrix of weakly exogenous variables; t is the linear trend restricted to the cointegration space and Dst is the matrix of the level-shift dummy (one in 1978 and another in 1994 to account for significant structural breaks in our long-run relationships).19 Γi is the unrestricted matrix of the coefficients in the short run and has a dimension of p × p, while ωi and θi are the coefficients of the variables that have been considered prior to analysing the weakly exogenous variables (Zt) and the level-shift dummy (Dst), respectively. Finally, the parameter (γDt) contains a vector of unrestricted dummy variables (a permanent dummy in 1976 and in 1989) and its corresponding coefficients. In addition, we assumed that the error term, εt, is an i.i.d. Gaussian sequence N(0, Ω) and the initial values, Yk+1,…Y0, are fixed. Table 1: Test for unit root Model 1 (trend& const) Model 2 (constant) Phillips-Perron Phillips-Perron Variable Levels Diff. Levels Diff. Lgdp -1.75 -5.04*** 0.73 -4.77*** Lprod -0.23 -6.47*** 5.47 -4.01*** Lrer -1.63 -5.43*** -1.01 -5.47*** LgdpUS -4.13 -4.89*** -1.76 -4.81*** Lrd -2.39 -6.40*** -0.19 -5.59*** Linv -3.14 -5.90*** -0.52 -5.82*** Lim -2.08 -5.24*** 0.94 -4.68*** ADF ADF Variable Levels Diff. Levels Diff. Lgdp -1.73 -4.44*** 0.25 -4.48*** Lprod -0.56 -4.62*** 2.30 -5.35*** Lrer -1.46 -5.50*** -1.01 -5.53*** LgdpUS -4.93* -4.90*** -1.39 -4.37*** Lrd -2.38 -4.60*** -0.43 -4.63*** Linv -4.88* -4.81*** -0.45 -4.88*** Lim -3.48 -5.28*** -0.48 -5.25***

Model 3 (none) Phillips-Perron Levels Diff. 22.38 -2.61** -3.11* -3.52*** 1.95 -4.94*** 10.50 -1.91*** 5.25 -4.12*** 10.30 -3.46*** 6.41 -2.86*** ADF Levels Diff. 11.01 -2.22** -2.69* -3.48*** 2.09 -4.93*** 10.26 -2.17*** 3.05 -4.14*** 5.87 -3.26*** 3.72 -2.99***

Note: **Rejection of the null hypothesis at 5%; *** Rejection of the null hypothesis at 10%

We start the analysis with a five-dimensional system that alternatively includes the GDP or labour productivity (output and labour productivity models, respectively), together with imports, investment, R&D expenditure and real exchange rate. We assumed that the US GDP is a weakly exogenous variable, and it was included as a control variable to capture the foreign economic conditions. The first step of this methodology is to identify the order of integration of the variables considered in our study. This analysis can be seen in Table 1, where, according to the Phillips-Perron and Augmented-Dickey-Fuller unit root tests, our variables are integrated of order one, and reject two unit roots on levels. Thus, any combination of non-stationary variables produces a stationary relationship if they are cointegrated. Moreover, in all the models estimated, two lags are enough to prevent autocorrelation problems and to capture the 19

We will now go on to explain the empirical relevance of these intervention dummies. The criterion to

include a dummy was (| εˆ 1 , t |> 3.3σˆ ε ) . For further details of the impact of deterministic components in the VAR Model, see Juselius (2007).

7

effects of dynamics, following the Hannan and Quinn criterion. Finally, both models were tested with the battery of miss-specification tests without showing any autocorrelation or normality problems.20 Once we have the well-specified model, we can test the rank of the long-run matrix. This can be done through the trace test, although it is not the only relevant information. We have complemented this test by looking at the root of the companion matrix, the magnitude and the significance of the alpha coefficients and the graphics of the cointegrated vectors, among other important information. For the sake of space limitations in this paper, we only report the analysis of the trace test and the roots of the companion matrix in order to identify the stationary long-run relationships. The procedure starts by investigating the null hypothesis r = 0, namely, absence of cointegration, and if it is rejected, we test the next null hypothesis r = 1 and so on until the hypothesis is not rejected. In accordance with the trace test and the roots of the companion matrix reported in Table 2 and 3, everything seems to indicate that in both the models considered, i.e. the productivity and the output model, there are two long-run relationships. Table 2: Determination of the Rank Test Productivity model (above) and output model (below) p-r r Trace Trace* 95% p-value p-value 2 0 104.27 93.38 54.20 0.000 0.000 1 1 40.86 38.65 27.78 0.001 0.002 p-r 2 1

r 0 1

Trace 99.24 41.11

Trace* 89058 39.33

95% 53.22 28.55

p-value 0.000 0.001

p-value 0.000 0.002

Table 3: Roots of the companion matrix of the productivity model (left) and output (right) Roots H(0) H(1) H(2) Roots H(0) H(1) H(2) 1 1 1 1 1 1 0.68 0.69 2 1 0.71 0.63 2 1 0.49 0.52 3 0.29 0.71 0.63 3 0.66 0.49 0.52 4 0.29 0.27 0.02 4 0.66 0.30 0.25 Note: (*) corresponds to the trace test with Bartlett’s correction. The asymptotic distributions have been simulated for the current deterministic specifications

In addition and parallel to the identification of the number of long-run relationships by the trace test, all the endogenous variables were tested for long-run exclusion, and the weak exogeneity test was applied to all variables of labour productivity and output models. Long-run exclusion was rejected in the case of investment, imports, R&D, output and productivity, and the weak exogeneity test shows that in both cases the only endogenous variables were output, productivity and investment.21

20

Available upon request from the authors. We applied the weak exogeneity test in the productivity model to show that the real exchange rate was weakly exogenous with a p-value of 0.05, together with imports and R&D expenditure, which had a p-value of 0.10 and 0.31 respectively. In the output model, the real exchange rate, imports and R&D expenditure were weakly exogenous with p-values of 0.29, 0.36 and 0.40 respectively.

21

8

4. Results

After determining the number of long-run relationships that exist in our two models and the endogenous and exogenous variables, the next step is to identify the cointegrated vectors in order to achieve economic interpretability of our relations. The procedure starts by testing the structural hypothesis on the non-significant coefficients ~ until the most irreducible form is reached. If Π = α β ’, the general linear hypothesis on β can be tested in the following way: H β : β = (H1ϕ1, …, Hi ϕi)

where Hi (p x (p-mi)) imposes mi restrictions on βi. The hypothesis is asymptotically distributed as χ2. The long-run economic relationships that were identified for labour productivity are as follows: (2.1)

lprod = 0.21linv + 0.24 lim+ 0.24 D s 94 [6.02]

[10.78]

[9.70]

(2.2)

linv = 0.31lrd + 0.20 D78 + 0.08t [5.53]

[9.92]

[33.17]

And for the output model:22 lg dp = 0.38linv + 0.42lrer + 0.19 lim+ 0.18D s 94 + 0.07 D s 78 [9.07]

[2.96]

[6.00]

linv = 0.24lrd + 0.24 D s 78 + 0.09t [3.49]

[8.38]

[5.36]

(3.1)

[2.11]

(3.2)

[35.36]

In equations (2.1) and (3.1), namely the first economic relation found in each model, we have normalized the long-run relationship in labour productivity and output, respectively, while in equations (2.2) and (3.2) the normalization was carried out in the investment variable. In both models, all coefficients are significant and show the expected signs. The stationarity of these relations cannot be rejected with a p-value of 0.30 for the productivity model and 0.37 for the output model. In accordance with the battery of stability tests, the concentrated version of the model is reasonably stable in all cases.23 22

In this model, although the statistic associated with Ds78 is relatively low, we allowed it in the first cointegrated vector given that an ambitious economic reform process began at the end of that year. 23 These tests are available from the authors.

9

Although we estimated two models, one for labour productivity and another for output, we found similar results. The first long-run relationship in each model (equations 2.1 and 3.1) shows that investment and imports positively influence labour productivity and output level in the long run, while the second relation found in these models (equations 2.2 and 3.2) shows that innovation activities and investment are to a certain extent complementary. These results are more in agreement with the hypothesis of endogenous growth models, which we outlined in section 2, than with the traditional models of growth. The long-run effect of imports and investment on labour productivity and output probably account for both the embodied technological progress associated with capital accumulation and the technology transfer from abroad that are associated with imports. In addition, gains in competitiveness accomplished through the real exchange rate were also seen to give rise to a positive effect on the determination of output level in the long-run. This last finding is in agreement with the development approach to currency management and has been considered a key factor to boost exports, income, employment and savings. The most important channels by which exchange rate levels affect long-run growth are related to investment and technological change.24 A similar conclusion has been reached by Rodrik (2007), who argues that the Chinese economy has made use of the continuous depreciations of its currency as an additional instrument of economic policy to enhance long-run growth. Interpreting the role played by the deterministic components included in the models is important to understand the stability of our long-run relationships and the role played by some exogenous reforms and shocks. The Ds78 captures the structural break in our relationships due to the shock associated with the beginning of the reform programmes at the end of 1978. Although “even after 1978, the pace of economic change was slow” (Bramall, 2000, p. 13), nobody questions the fact that the introduction of reform programmes in China to readjust the economy when Deng Xiaoping came to power at the end of the 70s signals the beginning of a new phase in Chinese economic development. The shock captured by Ds78 implies that after that year investment and output both increased more than the magnitude implied by the rest of the economic variables included in the relationships. This finding is also reasonable given that the process of decentralization and the increase in the presence of the non-state sector, as well as the start of the “open door” policy in China, all favoured the acceleration of output and investment, which in turn affected output and productivity. A similar effect is implied by the shift in relationships captured by Ds94. This dummy takes account of the positive effect on labour productivity and output, given that during the 90s a series of continuous reforms, such as the unification of the exchange rate or a tax reform, were implemented by the Chinese government. Finally, the linear trend in the investment relation could be associated with other determinants of investment that are not considered in our model. However, in order to complete the specification of the models that were estimated and to be more precise in the economic implications of the results, we identified the dynamics of labour productivity and output. The short-run dynamic adjustment structure was estimated by conditioning on the cointegration relations in equations 2.1, 2.2, 3.1 and 3.2, expressed in terms of error correction mechanisms (ecm). The results are reported in Table 4. To save space, the dummy variables are not included.

24

For more details, see Gala (2008).

10

Like the method used in the identification of the long-run structure, the nonsignificant coefficients are set to zero until the most irreducible form is reached. We test these restrictions on the coefficients by the Likelihood Ratio test, which is distributed as χ2 and, if not rejected, it implies that all the restrictions we have imposed are accepted by the data. Furthermore, it is possible to test the direction of the causality in the long and short run. The causality in the long run is established by the significance of the ecm, that is, our long-run relations, while the causality in the short run is analysed through the significance of the lags of our variables. Table 4: Dynamics of labour productivity and output (statistics in brackets)

∆lprodt-1

Labour Productivity model ∆lprod ∆linv 0.42 1.27 (5.48) (6.60)

∆lgdp t-1 0.37 (3.96) 0.04 Δlim (3.45) -0.03 Δlimt-1 (-2.85) -0.41 -1.07 Δlrer (-5.47) (-4.22) Δlrert-1 0.09 0.33 Δlrd (4.87) (5.46) 0.08 Δlrdt-1 (4.53) ecm1(t-1) -1.16 -1.53 (-8.45) (-2.01) -0.55 ecm2(t-1) -1.62 (-8.11) (-7.23) LR over-identifying restrict. short-run struct. χ2(15)=24.211 (0.0616) ∆lgdpusa

∆lgdp

GDP model ∆linv

0.29 0.77 (3.23) (3.04) 0.29 0.55 (2.18) (1.60) 0.05 (3.25) 0.17 (4.20) -0.33 -1.13 (-4.16) (-5.29) -0.16 -0.50 (-2.03) (-2.22) 0.12 0.27 (5.31) (4.35) 0.10 0.15 (3.84) (2.16) -0.52 -0.52 (-7.68) (-2.84) -0.44 -1.47 (-6.29) (-7.77) LR over-identifying restrict short run struct. χ2(9)=15.105 (0.0881)

Note: The long-run relations found in the dynamics of both the models are expressed as an error correction mechanism. ecm1 = lprod – 0.21linv - 0.24lim - 0.24Ds94 ecm2 = linv - 0.31lrd - 0.20Ds78 - 0.08t ecm1 = lgdp - 0.38linv - 0.19lim - 0.42lrer - 0.18Ds94 - 0.07Ds78 ecm2 = linv - 0.24lrd - 0.24Ds78 - 0.09t

labour productivity and output (2.1)’ (2.2)’ (3.1)’ (3.2)’

In the dynamics of the productivity model, it can be observed how the productivity equation adjusts towards equilibrium with the first long-run relationship that is found (ecm1), while the investment equation is error correcting with the second (ecm2). In both cases the alpha coefficients are negative and statistically significant. The speed of adjustment towards the long-run equilibrium in both relations is fast and around six months. Similar interpretations are possible with the output model. Our main findings are consistent with an import-led growth effect and show a positive relation between imports and labour productivity/output. The causality runs in one

11

direction from imports to productivity/output, as can be seen in the coefficients of the error correction mechanism, which are highly significantly negative. We did not find causality running in two directions between imports and productivity/output given that imports become weakly exogenous in our estimated models. The implication of weak exogeneity of one or more variables is that they influence the long-run stochastic path of the other variables of the system, although at the same time they are not influenced by the other variables. In addition, we found an investment-led growth effect that was probably due to embodied technological progress associated with capital accumulation; in other words, it seems that supply factors predominate among the determinants of capital accumulation in the Chinese economy (Madsen, 2002). Thus, according to our findings, capital accumulation is not the only source of growth – there is also technological progress, in this case, in the form of embodied technological progress in capital goods and imports. We did not find that the increase in investment causes imports directly, as Rodrik (1995) suggests, because his hypothesis claims that imports are endogenously determined by investment, and the direction of the causality runs from investment to imports. The increase in imports then encourages exports and hence economic growth. However, according to our estimations, imports become exogenous and they are not influenced by investment. Our results imply that investment and imports account for labour productivity and output directly in the long run. Additionally, we found that R&D expenditure directly stimulates investment and has an indirect positive effect on productivity and output. This result, which is robust to different specifications, is interesting in the sense that R&D has two faces. On the one hand, it generates new innovations and facilitates the assimilation of these new discoveries (Griffith et al., 2004; Cameron et al., 2005). And on the other hand, in accordance with Howitt and Aghion (1998), innovation activities and capital accumulation are complementary and could play a critical role in long-run growth. Thus, our results are more in agreement with the conclusions drawn by these authors. In the short-run dynamics, we can find common effects in both the models that were estimated. First, our results indicate that the real exchange rate negatively affects productivity/output and investment growth, which can probably be explained by the deterioration of terms of trade. Second, innovation activities measured by R&D expenditure have played a key role in the process of growth and stimulate investment. Third, we find a procyclical effect of foreign demand that probably captures the influence of the international business cycle. Finally, we found that the increase in imports stimulates investment in the output model.

Comments and Conclusions

In the empirical literature, capital accumulation and export promotion policy have been considered the main forces driving the rapid growth of China. However, the endogenous growth theory highlights the role played by imports rather than exports in the process of economic growth. Although there is theoretical support for the relationship between imports and economic development, to the best of our knowledge there is no empirical evidence for the case of China. This study has attempted to cover this gap. We investigated the effects of imports and investment together with innovation activities on labour productivity and output for the period 1964-2004. Our results

12

indicate that imports and investment have positively influenced labour productivity and output in the long run. These findings are consistent with the import-led growth hypothesis, the causality running in one direction from imports to growth. Moreover, we found an investment-led growth effect where the causality again runs in one direction from investment to economic activity. These results indicate that both factors affect economic development and in this sense partially contradict the arguments put forward by Rodrik, who supported the hypothesis that trade is more a consequence of the rapid growth in Asian countries than a cause. Finally, we found some complementarities between imports and investment, as the main factors of long-run growth in China, and between R&D expenditure and investment. According to our results, the Chinese economic policy that was applied seems to have been an option with successful results. This strategy has been based on acquiring advanced technology from abroad. The majority of imports in China are machinery, equipment and intermediate goods, which are relatively cheaper and easier to implement in the production process, thereby improving efficiency and productivity. Thus, our results for the Chinese economy are important because, as in the case of other developing countries, China has no comparative advantage in the production of capitalintensive goods. In this case, access to cheaper capital imports and intermediate goods becomes crucial to the development of domestic technological capabilities through assimilating and adapting the advanced technology transferred from R&D intensive countries. However, this policy has not been entirely free of criticism. Although it seems undeniable that it has facilitated the industrialization process since the fifties, doubts have been raised about the degree of assimilation of the technology imports in the Chinese economy, given the lower level of workers’ skills and the ability to imitate or innovate on foreign technology. On this issue, China has made considerable improvements, but it is still a long way from the competitive foreign markets and is one of the aspects that should be changed in order to guarantee the success of this strategy of growth. In addition, although the open-door policy has integrated China into the global market, the exchange rate policy has kept its currency undervalued with regard to its hypothetical equilibrium. This in turn leads to imports becoming more expensive and the creation of foreign exchange limitations, given the interventions in the financial market in the greater part of the period considered. Thus, exports were promoted to mitigate the foreign exchange constraints. Economists and politicians have questioned the extent to which China can sustain this economic policy. Critics argue that the export promotion policy is unsustainable because foreign countries cannot afford the large amount of products that has to move into and out of China. However, others argue that although China is the fourth largest exporter in the world, about half of China’s exports are goods that are processed with imported materials. If such processing trade is deducted, the opening up of China is still far lower than the global level (Hu, 2007). Finally, the sustainability of the investment-to-GDP ratio has also been subject to debate in the literature. China has followed a deliberate strategy of stimulating saving and investment. This policy has meant that China, today, is the country with one of the largest ratios of investment to output. For most of the period analysed, the investment growth rate is greater than the GDP growth rate. These continuous increases in investment have led to a decrease in capital productivity. This is one of the arguments that many authors have used to justify alternatives in the sources of growth in China.

13

Others, however, argue that a look at the cross-sectional distribution of investment across Chinese provinces shows it to be concentrated on the coast, while the investment-to-GDP ratio remains at lower levels in the central and western provinces. In this regard, it is advisable to continue these investment efforts if they are to be redistributed across provinces, especially in central and western regions. Finally, the high levels of savings and investment imply lower levels of consumption in the Chinese economy. This characteristic, which is common to Asian countries, could cause negative effects on economic development, if one of the components of the internal demand fails to stimulate growth. To sum up, although China has experienced an impressive amount of progress in developing its economy, and the openness strategy has provided access to better technology and more and better intermediate inputs, it is the domestic effort to learn and assimilate new technologies that determines the degree of development of the economy. In this sense, more reforms are needed, especially with regard to improving efficiency in allocating economic resources and the level of education. References

Amiti, M. and Konings, J. (2005), ‘Trade liberalization, intermediate inputs and productivity: Evidence from Indonesia.’ CEPR 5104. Awokuse, T.O. (2007), ‘Causality between exports, imports, and economic growth: Evidence from transition economies.’ Economics Letters 94, pp. 389-395. Awokuse, T.O. (2008), ‘Trade openness and economic growth: is growth exportled or import-led?’ Applied Economics 40, pp. 161-173. Bai, C., Hsieh, C. and Qian, Y. (2006), ‘The Return to Capital in China.’ NBER 12755 Bramall, C., (2000), ‘Sources of Chinese Economic Growth 1978-1996.’ Studies on Contemporary China. Oxford University Press. Cameron, G., Proudman, J. and Redding, S. (2005), ‘Technological convergence, R&D, trade and productivity Growth.’ European Economic Review 49, pp. 775-807. Caselli, F. and Wilson, D.J., (2004), ‘Importing technology.’ Journal of Monetary Economics 51, pp. 1-32. Chow, G. (1987), ‘Money and price determination in China.’ Journal of Comparative Economics 11, pp. 319-333. Chow, G., (1993), ‘Capital Formation and Economic Growth in China.’ The Quarterly Journal of Economics 108, pp. 809-842. Chow, G., (2004), ‘Knowing China.’ Singapore: World Scientific Publishing Co. Chow, G. (2006), ‘Are Chinese Statistics Reliable?’ Cesifo Economic Studies 52, pp. 396-414. Coe, D.T, Helpman, E. and Hoffmaister, A.W. (1997), ‘North-South R&D Spillovers.’ The Economic Journal 107 pp. 134- 149. Conroy R. (1986) ‘China’s technology import policy’ The Australian Journal of Chinese affairs 15, pp. 19-56.

14

Eaton, J. and Kortum, S., (2001), ‘Trade in capital goods.’ European Economic Review 45, pp. 1195-1235. Gala, P. (2008), ‘Real Exchange Rate Levels and Economic Development: Theoretical Analysis and Empirical Evidence.’ Cambridge Journal of Economics 32, pp. 273-288. Griffith, R., Redding, S, and. Reenen, J.V. (2004), ‘Mapping the two faces of R&D: Productivity Growth in a Panel of OECD Industries.’ The Review of Economics and Statistics 86, pp. 883-895. Grossman, G.M. and Helpman, E. (1991), ‘Innovation and Growth in the Global Economy.’ Cambridge, Massachusetts, MIT Press. Howitt, P. and Aghion, P. (1998), ‘Capital Accumulation and Innovation as Complementary Factors in Long-Run Growth.’ Journal of Economic Growth. 3, pp. 111-130. Holz, C.A. (2005), ‘OECD – China Governance Project: The Institutional Arrangements for the Production of Statistics’, OECD Statistics Working Papers, 2005/1, OECD Publishing. Hu, A. (2007) ‘Economic and Social Transformation in China. Challenges and Opportunities.’ Routledge Studies on the Chinese Economy. Johansen, S. (1988) ‘Statistical Analysis of Cointegration Vectors.’ Journal of Economic Dynamics and Control 12, pp. 231-254. Johansen, S. and Juselius, K. (1990) ‘Maximum Likelihood Estimation and inference on cointegration-with Applications to the Demand for money.’ Oxford Bulletin of Economics and Statistics 52, pp. 169-210. Johansen, S. (1995), ‘Likelihood-Based inference in cointegrated vector autoregressive models.’ Advance Texts in Econometrics. Oxford University Press. Juselius, K. (2007), ‘The cointegrated VAR model: Econometric methodology and macroeconomics applications.’ Oxford University Press. Krishna, K., Ataman, O. and Norman, R. S, (2003) ‘Trade, Investment and Growth: Nexus, Analysis and Prognosis.’ Journal of Development Economics, 70, pp. 479-499. Lee, J.W.(1995) ‘Capital goods imports and long-run growth.’ Journal of Development Economics 48, pp. 91-110. Li, X. (2000), ‘The Great Leap Forward, Economic Reforms, and the Unit Root Hypothesis: Testing for Breaking Trend Functions in China’s GDP Data.’ Journal of Comparative Economics 28, pp. 814–827. López R.A.(2006), ‘Imports of intermediate inputs and plant survival.’ Economics Letters 92, pp. 58-62. MacDonald, J. M. (1994), ‘Does import competition force efficient production?’ The Review of Economic and Statistics 76, pp. 721-727. Madsen, J.B. (2002), ‘The causality between investment and economic growth.’ Economic Letters 74, pp. 157-163. Mazumdar, J. (2001), ‘Imported machinery and growth in LDCs.’ Journal of Development Economics 65, pp. 209-224.

15

Narayan, P. K. and Sun, G. Z., (2007). ‘The Division of Labor, Capital, Communication Technology and Economic Growth: The Case of China 195299.’ Review of Development Economics 11, pp. 645-664. Nielsen, H., (2004). ‘Cointegration analysis in the presence of outliers.’ Econometrics Journal 7, pp. 249-271. Phylaktis, K. and Girardin, E. (2001), ‘Foreign exchange markets in transition economies: China.’ Journal of Development Economics 64, pp.215-235. Rivera-Batiz L.A. and Romer P.M., (1991), ‘Economic Integration and Endogenous Growth.’ Quarterly Journal of Economics 106, pp. 531-556. Rodrik, D. (1995), ‘Getting interventions right how South Korea and Taiwan grew rich.’ Economic Policy 10, pp. 53-107. Rodrik, D. (2007), ‘The Real Exchange Rate and Economic Growth: Theory and Evidence. Undervaluation is good for growth, but why?’ Working paper. Harvard University. Shi, Y. (1998), ‘Chinese firms and Technology in the Reform Era.’ Routledge Studies in the Growth Economies of Asia. Siebert, H. (2007), ‘China: Coming to Grips with the new global player.’ The World Economy 30, pp. 893-922. Thangavelu, S. M. and Rajaguru, G. (2004), ‘Is there an export or import-led productivity growth in rapidly developing Asian countries? A multivariate VAR analysis.’ Applied Economics 36, pp. 1083-1093. Traca, D.A. (2002), ‘Imports as competitive discipline: the role of the productivity gap.’ Journal of Development Economics 69, pp. 1-21. Wall, D. (1968), ‘Import Capacity, import and economic growth’. Economica 35, pp. 157-168. World Bank. 1993. ‘The East Asian Miracle’, a World Bank Policy Research Report, Oxford: Oxford University Press. Yao, S. (2000) ‘How important is agriculture in China’s economic growth?’ Oxford Development Studies 28, pp. 33-50.

16