The relationship between trade openness and growth is a highly debated topic in the growth and development literature, yet this issue is far from being resolved. There is a long history of research, both theoretical and empirical, that provides at least an answer to the question: 'Does openness to trade result in the growth of output (say, GDP)?' But currently there is no consensus, either empirically or theoretically, on the nature of the relationship between trade openness and output growth. In part, this is because the mechanisms behind it are not well understood.
The existing empirical literature however does not provide clear evidence on relationship between trade openness and growth. Many studies provide evidence that increasing openness has a positive effect on GDP growth. On the other hand some studies report that it is difficult to find robust positive relationships or even that there is negative relationship between openness and growth. Some studies, among others Rodriguez and Rodrik (1999) and Rodriguez (2006), critically argue that trade policy variables are mostly uncorrelated with growth, while the trade shares can correlate with income levels and growth rates. But the complexity of links of causality and endogeneity among trade shares, growth and other sources of growth make a difficulty to define a strong effect of openness on economic growth. Theoretical growth studies suggest very complex and different relationships between openness and growth and the empirical evidence is not unambiguous. The new growth theory supposes that "a country's openness to world trade improves domestic technology, and hence an open economy grows faster than a closed economy through its impact on technological enhancement" (Jin, 2006). Harrison (1996) asserted that openness to trade provides access to imported inputs, which embody new technology, increases the size of the market faced by the domestic producers, which raises the returns to innovation, and facilitates a country's specialization in research-intensive production.
In line with potential dynamic gains of trade openness, most early empirical studies have examined a set of trade openness measures and their correlation with each other and with economic growth and found a clear positive link. For example, Harrison (1996) looked at a number of openness indicators that turned out to have a positive 'association' with economic growth and produced evidence in support of bi-directional causality between openness (trade share) and economic growth. Recent research, however, has questioned the robustness of the relationship. For instance, Harrison and Hanson (1999) show that the often quoted Sachs and Warner (1995) findings do not provide evidence for an openness and growth link as claimed. Rodriguez and Rodrik (1999) confirm the Harrison-Hanson critique and argued that much of the work to correlate trade openness and economic growth has been plagued with subjective and collinear measures of openness that, though positively related with economic growth, arrive at their conclusions through problematic econometric methodologies. Harrison (1996) and Pritchett (1996) show that the various measures of trade openness tend to be only weakly correlated and are often of the wrong sign.
In general, empirical studies suffer from a number of shortcomings, and as a result they have not resolved the questions surrounding the correlation between openness and growth. Baldwin (2000) offers explanations for the differences among researchers of the openness-growth nexus. According to him, while econometric analyses based on quantitative data are limited by the scope and comparability of available quantitative data, differences in what investigators regard as appropriate econometric models and tests for sensitivity of the results to alternative specifications that may be based in part on the personal policy predilections of authors and can also result in significant differences in the conclusions reached under such quantitative approaches. If these studies used measures that were even slightly correlated, then the empirical literature together could be taken as proof of a positive relation between openness and growth.
Baliamoune-Lutz and Ndikumana (2007) observed that, from a methodological standpoint, the weak link between trade liberalization and growth may be attributed to measurement imperfections: the indicators used in empirical analysis may not capture the true essence of openness. Indeed, due to lack of data on indicators of trade openness as a policy, empirical studies (as this one does) resort to measures of trade outcomes, i.e., trade volume, as proxies for trade openness. It is assumed that positive trade outcomes are an indication of a policy environment that is at least not anti-trade. Moreover, a high trade volume indicates exposure to international markets with the associated benefits (e.g., technological transfer), which openness policies seek to achieve. Thus, to some extent trade outcomes do carry some indication of the effects of trade liberalization. Nonetheless, results from analyses using trade volume as a measure of trade openness have to be interpreted cautiously. Indeed, variations in the volume of trade do not always reflect actual government policies that promote or hinder trade. For instance, fluctuations in commodity prices result in changes in trade flows even in the absence of shifts in trade policy.
The weak empirical evidence on the link between trade liberalization and growth can also be due to problems of misspecification. In particular, the effects of trade liberalization may materialize only with a lag. In the short run, liberalization may have negative effects, especially by undermining domestic production because of competitive imports, retarding growth (Mukhopadhyay 1999). Hence, to the extent that these negative short-run effects and the expected delayed positive effects occur consecutively, growth would exhibit a J-curve type of response to trade openness (Greenaway et al. 2002). Therefore, empirical studies may yield inconclusive and even misleading results if these dynamic and counterbalancing effects are not fully taken into account.
Another explanation relates to the structure of trade. Whether a country benefits from trade liberalization or not in terms of growth depends on the composition of trade. Mazumdar (1996) hypothesized that the composition of trade determines the strength of the "engine of growth." Indeed, Lewer and Van Den Berg (2003) find evidence supporting the view that countries that import capital goods and export consumer goods grow faster than those that export capital goods. The evidence suggests that African countries and developing countries in general would benefit from trade most by promoting exports of labor-intensive goods and services while encouraging imports of capital goods (Lopez 1991). This implies that the current export boom which is driven by capital-intensive sectors such as oil is not likely to generate growth that is sustainable, especially because of the low gains in employment creation and limited spillover effects on non-oil sectors.
Dollar (1992) brought an important contribution to the trade and growth debate. The author defines openness as the combination of two dimensions: (i) a low level of protection, hence of trade distortions and (ii) a stable real exchange rate so that incentives remain constant over time. From that very definition, follow two measures openness: a trade distortion index, and a real exchange rate variability index. The distortion index measures the deviations from the Law of One Price after controlling for the impact of nontradables. The variability index captures the variance of the real exchange rate. The author considers a sample of 95 countries over the period 1976- 1985 and regresses average per capita growth upon his openness indexes and the average investment rate. Both the distortion index and the variability index are significantly negatively correlated with growth and the investment rate comes out with a significantly positive coefficient.
Dowrick (1994) tests whether trade openness affects output growth and/or investment. He considers a sample of 74 countries over the period 1960-1990. As openness indicator, the author considers the residuals of an OLS cross-country regression of the average trade intensity upon a constant and average population. In a second stage, the author runs cross-country OLS regressions of average per capita GDP growth upon the average investment rate, the initial GDP level and his openness indicator. The coefficient on openness is significant and positive. Moreover, dropping the investment rate considerably lowers the overall fit of the model but enhances the coefficient on openness, which, according to the author "suggests that openness works partly through increased investment rates". In a third stage, the author computes decade averages for his variables and turns to panel data techniques, arguing that such techniques "enable some control for time-invariant country-specific factors such as institutional arrangements that might be correlated with the explanatory variables". The author uses labour productivity growth as dependent variable and estimates both fixed-effects and random-effects models. He reports that the coefficient on openness is still significant and positive, but its point estimate is much lower than in the OLS specification. In a fourth set of regressions, the author also considers growth in openness instead of openness itself. The impact of that variable on growth is still significantly positive as far as developing countries are concerned, but becomes insignificant when turning to the sample of developed countries. The author interprets this as reflecting the fact that "static efficiency effects of trade liberalization are negligible for countries with well-developed markets." Finally, in its Conclusions, the author cautions that his results, showing the beneficial effects of increased openness, hold on average, but are not a universal truth, valid always and everywhere. In particular, he stresses that "trade liberalization can indeed stimulate growth in the aggregate world economy. Whilst trade may have such positive effects for some countries, it may conversely lock in other countries into a pattern of specialization in low-skill, low-growth activities".
Sachs and Warner (1995) brought a seminal contribution to that literature. Their central hypothesis is that some developing countries fail to grow rapidly enough as to converge because they are simply not open to trade. In their own words: "convergence can be achieved by all countries, even those with low initial level of skill, as long as they are open and integrated in the world economy". To check their hypothesis, the authors first carefully build and discuss an openness measure. Building upon a sample of 135 countries over the period 1970-1990, they construct an openness dummy variable that is zero if any of the 5 following conditions is true:
non-tariff barriers covering 40% or more of trade,
average tariff rate above 40%,
black market premium above 20%,
the economy is ruled by a socialist system, or
there is a state monopoly on exports.
Otherwise, if none of these 5 conditions is fulfilled, the openness dummy is one. The authors first divide their countries sample into open ones and closed ones, and show that closed countries have grown at about the same rate (essentially about 0.7% a year), no matter whether they are developed or not. By contrast, open developing countries have grown much faster than their developed counterparts (4.49% versus 2.29%). Going beyond these stylized facts, the authors re-do the same regressions as in Barro (1991) and add their openness dummy to them. Without the dummy, the results are sensibly the same as in Barro (1991). After adding the openness dummy in the regressors list, it appears its coefficient is highly significant. The point estimates suggest that open economies grow on average 2.45% faster than closed ones. Moreover, educational attainment variables become even less significant than in Barro (1991), which leads the authors to think that "...growth rate over this period was determined less by initial human capital levels than by policy choices". They also address a specialization-related issue. Specifically, they test whether trade openness condemns raw materials exporters to non-industrialisation and whether closed trade promotes industrial exports in the long run. To do this, they regress the change in the share of primary exports on openness. They find that "open economies tend to export more rapidly from being primary-intensive to manufactures-intensive exporters. The difference in speed of adjustment is statistically significant".
Harrison (1996) starts from the judgment that "it should be evident that no independent measure of so-called 'openness' is free from methodological problem". Therefore, to make her point, she collects as many different openness indicators as she can, about 7 of them, and she checks the consistency of the results across all these indicators. She uses various samples, whose time spans range from 1960-1988 to 1978-1987, and the country coverage varies from 51 to 17. She first runs typical cross-country growth regressions. It appears that only one measure of openness out of 7, namely the black market premium, has a significant impact on growth. To explain this weak result, the author argues that a pure cross-section specification, based upon long-run averages, is not an adequate one. Indeed, though the use of long run averages appears as the most natural way to capture the determinants of long-run growth, they may also hide significant variations in individual countries' performances and policies over time. To test this idea, the author re-does her regressions using annual data for the same variables. She uses a panel fixed-effects specification to take into account unobserved country specific differences in growth rates. Results show a stronger link between openness and growth since 3 indicators become significant at the conventional 5% level. The author next argues that such a yearly frequency is too high if one is interested in long-run growth, since results may be affected by short-term, conjectural, variations. She therefore considers a third - "intermediate" - specification, based on five-year averages and reports that, again, 3 indicators come out with a significant coefficient. The message from these results, as the author states, is that "the choice of the time period for analysis is critical". However, an interesting regularity appears across all specifications: when openness is significant, it is always in the sense that greater openness is associated with higher growth.
Edwards (1998) also uses an important number of openness indexes to investigate the trade and growth relationship. He considers a sample of 93 advanced and developing countries, and estimates a growth equation with a panel data random effects model. From that model, he computes factor shares, which are then used to get TFP estimates. Concentrating on a cross-section of 1980s averages, TFP growth is finally regressed upon initial income level, initial human capital level, and no less than 9 openness indicators, each one of them in turn. The author reports that "in all but one of the 18 equations the estimated coefficient on the openness indicator has the expected sign and in the vast majority of cases it is significant". Moreover, the coefficient on initial human capital is always significant and positive. Regarding the initial income level, the coefficient is always negative and in 16 cases out of 18, it is significant though very low, which can be interpreted as evidence in favour of conditional convergence. To summarize, the authors concludes that his results "are quite remarkable, suggesting with tremendous consistency that there is a significantly positive relationship between trade openness and growth".
An important paper that is able to cast serious doubts about the consistency of the trade-growth relationship is the one by Rodriguez and Rodrik (1999). These authors consider a series of previous research results, among which Dollar (1992), Sachs and Warner (1995), and Edwards (1998). They re-do the computations in these papers, but slightly change the specifications (through the addition of some dummies, e.g.), add newly available data to the sample, or slightly change the estimation methods. They are able to demonstrate a fundamental lack of robustness of the results in the paper they review.
Frankel and Romer (1999) claim that openness, as measured by the ratio of total trade to GDP, should not be used as explanatory variable in the growth regressions. The trade ratio, the authors argue, is endogenous, and needs to be instrumented. To construct their instrument, the authors first argue that "as the literature on the gravity model of trade demonstrates, geography is a powerful determinant of bilateral trade". And they claim this is also true for total trade. Moreover, geography is completely exogenous. Therefore, the authors consider a database of bilateral trade between 63 countries for 1985 and they regress bilateral trade upon purely geographical indicators. For each country, the fitted values of trade are aggregated over all partners, and this aggregate is finally turned into an "ideal" trade share that can be used as an instrument for the observed one. The authors then estimate growth equations for a cross-section of 150 countries in 1985. They report a substantial impact of trade openness on income growth: increasing the trade share by 1% should raise income by between 0.5% and 2%. These findings are robust to various changes in specifications. The results also suggest that, controlling for openness, larger countries tend to experience higher growth rates, which could simply reflect that citizens living in larger countries engage more in within-country trade.
Baldwin and Sbergami (2000) argue that the reason why researchers failed to find a robust relationship between trade and openness is because that relationship is fundamentally nonlinear and non-monotonic. They raise the point that the fundamental engine of growth is human and physical accumulation, and that the link between capital accumulation and trade barriers is, in nearly all models, nonlinear and often even non-monotonic. They provide a formal 2Ã-2Ã-2 dynamic model with imperfect competition that gives rise to (i) a U-shaped relationship between ad-valorem tariffs and growth and (ii) a bell-shaped relationship between specific tariffs and growth. This model is then confronted to the data, i.e. for a variety of openness indicators (actually, 10 of them are considered), a quadratic model is estimated. It turns out that, in this new specification, for 6 of the 10 proxies both the linear and the quadratic terms are significant individually. The authors conclude that: "allowing for non-linearity does have a big empirical impact".
A number of other studies have looked at the relationship between average tariff rates and growth. Lee (1993), Harrison (1996) and Edwards (1998) found a negative relationship between the tariff rates and growth. The studies of Edwards (1992), Sala-i-Martin (1997) and Clemens and Williamson (2001) concluded that the relationship is weak. Rodriguez and Rodrik (1999) tried to replicate the result of Edwards (1998) and found that average tariff rates had a positive and significant relationship with total factor productivity (TFP) growth for a sample of 43 countries over the period 1980-1990.
In a recent study Yanikkaya (2003) used a large number of openness measures for a cross-section of countries over the last three decades. His analysis found a significant positive correlation between trade shares and growth. However, this study observed that different measures of trade barriers are positively associated with growth in the less developed countries. In recent empirical studies, one or more of the following indicators of openness in the table below are used:
Measure
Definition
Trade dependency Ratio
The ratio of exports and imports to GDP
Growth Rate of Exports
The growth rate of exports over the specified period
Tariff Averages
A simple or trade-weighted average of tariff levels
Collected Tariff Ratios
The ratio of Tariff revenues to imports
Coverage of Quantitative
Restrictions
The percentage of goods covered by quantitative restrictions
Black Market Premium
The black market premium for foreign exchange, a proxy for the overall degree of external sector distortions
Trade Bias Index
The extent to which policy increases the ratio of importable
goods' prices relative to exportable goods prices compared to the same ratio in world markets
Sachs and Warner Index
A composite index that uses several trade-related indicators;
tariffs, quota coverage, black market premiums, social organization and the existence of export marketing boards
Leamer's openness index
An index that estimates the difference between the actual trade flows and those that was expected from a theoretical trade model
Table : Openness indicators
(Rodriguez and Rodrik, 2000; Ogujiuba, Oji and Adenuga, 2004)
Grossman and Helpman (1991) and Matsuyama (1992) provide theoretical models where a technologically backward country specializes in a non-dynamic sector as result of openness, thus losing out from the benefits of increasing returns. Underlying this result, there is an imperfection in contracts or in financial markets that makes people obeys a myopic notion of comparative advantage.
Dollar and Kraay (2004) and Loayza, Fajnzylber, and Calderón (2005) run growth regressions on panel data of large samples of countries. Both papers use openness indicators based on trade volumes and control for their joint endogeneity and correlation with country-specific factors through GMM methods that involve taking differences of data and instruments. This implies that, although they continue to use cross-country data, these papers favor within-country changes as the main source of relevant variation. Both papers conclude that opening the economy to international trade brings about significant growth improvements. Wacziarg and Welch (2003) arrive to a similar, though more nuanced, conclusion from a methodologically different standpoint. Using an event-study methodology --where the event is defined as the year of substantial trade policy liberalization--, they find that liberalizing countries tend to experience significantly higher volumes of trade, investment rates, and, most importantly, growth rates. However, in an examination of 13 country-case studies, Wacziarg and Welch find noticeable heterogeneity in the growth response to trade liberalization. Although their small sample does not allow for definite conclusions, it appears that the growth response after liberalization is positively related to conditions of political stability.
Also, various empirical literatures offer some examples of non-linear specifications considering interaction effects. On the related topic of foreign direct investment, Borensztein, De Gregorio and Lee (1998) find that the growth effect of FDI is significantly positive only when the host country has, respectively, sufficiently high human capital and financial depth. Specifically in the analysis of growth effects of trade openness, an important antecedent of our work is the empirical study by Bolaky and Freund (2004). Using cross-country regressions in levels and changes of per capita GDP and controlling for simultaneity via external instruments, they find that trade opening promotes economic growth only in countries that are not excessively regulated. They argue that in highly regulated countries, growth does not accompany trade openness because resources are prevented from flowing to the most productive sectors and firms, and trade is likely to occur in goods where comparative advantage is actually missing.
Calderón, Loayza, and Schmidt-Hebbel (2004) interact in their panel growth regressions a measure of openness (volume of trade / GDP) with linear and quadratic terms of GDP per capita, which they regard as proxy for overall development. They find that the growth effect of trade opening is nearly zero for low levels of per capita GDP, increases at a decreasing rate as income rises, and reaches a maximum at high levels of income.
Chang, Kaltani and Loayza (2005) study how the effect of trade openness on economic growth depends on complementary reforms that help a country take advantage of international competition. They presented some panel evidence on how the growth effect of openness depends on a variety of structural characteristics. They use a non-linear growth regression specification that interacts a proxy of trade openness with proxies of educational investment, financial depth, inflation stabilization, public infrastructure, governance, labor-market flexibility, ease of firm entry, and ease of firm exit. They find that the growth effects of openness are positive and economically significant if certain complementary reforms are undertaken.
Giles and Stroomer (2005) develop flexible techniques for measuring the speed of output convergence between countries when such convergence may be of an unknown non-linear form. They then calculate these convergence speeds for various countries, in terms of half-lives, using a time-series data-set for 88 countries. These calculations are based on both nonparametric kernel regression and 'fuzzy' regression, and the results are compared with more restrictive estimates based on the assumption of linear convergence. The calculated half-lives are regressed, again in various flexible ways, on cross-section data for the degree of openness to trade. They find evidence that favours the hypothesis that increased trade openness is associated with a faster rate of convergence in output between countries.
On research studies that relate to Africa and Nigeria in specific, Sarkar (2007) examines the relationship between openness (trade-GDP ratio) and growth. The cross-country panel data analysis of a sample 51 countries of the South during 1981-2002 shows that for only 11 rich and highly trade-dependent countries a higher real growth is associated with a higher trade share. Time series study of individual country experiences shows that the majority of the countries covered in the sample including the East Asian countries experienced no positive long-term relationship between openness and growth during 1961-2002. He finds that the experience of various regions and groups shows that only the Middle Income group exhibited a positive long-term relationship.
Also, Baliamoune-Lutz and Ndikumana (2007) explore the argument that one of the causes for the limited growth effects of trade openness in Africa may be the weakness of institutions. They also control for several major factors and, in particular, for export diversification, using a newly developed dataset on Africa. Results from Arellano-Bond GMM estimations on panel data from African countries show that institutions play an important role in enhancing the growth effects of trade. They find that the joint effect of institutions and trade has a U-shape, suggesting that as openness to trade reaches high levels, institutions play a critical role in harnessing the trade-led engine of growth. The results from this paper are informative about the missing link between trade liberalization and growth in the case of African countries.
Likewise, Ogujiuba, Oji and Adenuga (2004) test the validity of trade openness for Nigeria's Long-Run Growth using a cointegration approach. They preferred the VAR approach for some reasons and their econometric results show that there is no significant relationship between openness and economic growth, and that unbridled openness could have deleterious implications for growth of local industries, the real sector and government revenue.
Moreover, Addison and Wodon (2007) study the macroeconomic volatility, private investment, growth, and poverty in Nigeria. Using cross-sectional data for 87 countries, they show that real per-capita growth over the period 1980-1994 was a function of productivity growth and investment rates, both of which were negatively effected by volatility (in terms of trade, real exchange rate, and public investments). When comparing Nigeria to high growth nations, they find that most of the growth differential can be attributed to Nigeria's higher macroeconomic volatility. Simulations suggest that if Nigeria had had lower levels of volatility and better macroeconomic policies, poverty would have been much lower than observed.
Nwafor (undated) examines the effects reduction of import tariffs will have on poverty in Nigeria. Using information on Nigeria's past experience with trade liberalisation, he examined the possible impacts on the economy with a view to making the reductions pro-poor.
Kandiero and Chitiga (2003) investigate the impact of openness to trade on the FDI inflow to Africa. Specifically, in addition to economy wide trade openness, they analyse the impact on FDI of openness in manufactured goods, primary commodities and services. The empirical work is conducted using cross-country data comprising of African countries observed over four periods: 1980-1985, 1985-1990, 1990-1995 and 1995-2001. They find that FDI to GDP ratio responds well to increased openness in the whole economy and in the services sector in particular.
Finally, Njikam, Binam and Tachi (2006) assessing the factors behind differences in total factor productivity (TFP) across SSA countries over the period 1965-2000. The cross-section, fixed-effects using annual data, fixed-effects using data in 3-year averages as well as the seemingly unrelated regression (SUR) results show that (i) openness to world trade is conducive to TFP in SSA region only if issues related to supply conditions such as poor transport and communication infrastructure, erratic supply of electric energy, corruption and bad governance, insufficient education of the labour force, etc. are adequately addressed, (ii) physical capital accumulation is important for TFP, (iii) the size of the financial sector matters for TFP, and (iv) population growth is conducive for TFP in some SSA countries and negative for TFP in other SSA countries.
2.4 Limitation of Previous Studies
The literatures of the previous studies are plagued with a lot of problems. First of all, it is worthwhile to note that the theoretical growth literature has given more attention to the relationship between trade policies and growth rather than the relationship between trade volumes and growth. Therefore, the conclusion about the relationship between trade barriers and growth cannot be directly applied to the effects of changes in trade volumes on growth. There was also no consensus on the nature of the relationship and nature of linear association (correlation) between openness and growth. Likewise, there is no generally accepted measure of openness, each study uses any of the openness indicators as it suit and please the researcher(s). Moreover, many of the existing empirical literatures are not country-specific, that is they deal with cross-sectional analysis, thus they did not provide for differential in nature and structure of various economies. Hence, developing countries, like Nigeria, are recommended policies which are based on research conducted for industrially advanced countries or even mixture of both.
2.5 Chapter Summary and Prospect
In this chapter, we have critically reviewed the relevant literature on the concept of openness and growth. Under the theoretical literature, we analysed the two arguments concerning openness and their various policy options as well as benefits and problems of openness to trade. We also reviewed handful empirical studies conducted by various researchers and their conclusions. Finally, we examined the fundamental and methodological limitations facing the various empirical studies reviewed. In the next chapter, we are going to examine the analytical framework and methodology of this research work. We will also examine the various criteria which will be used to evaluate the significance and accuracy of the research results and the methodology adopted.
CHAPTER THREE: METHODOLOGY
3.1 Analytical Framework
The primary aim of every economic research is to arrive at a conjunction of economic theory, actual measurement using the theory and techniques of statistical inference as the matching bridge (Haavelmo, 1994). The economic theory makes statements or postulates hypotheses that are mostly quantitative (and in some cases qualitative) in nature and as such, it is the choice of the modeler or the researcher to validate these hypotheses using appropriate models in line with current development and befitting method of estimation and inference.
Economic theory and some empirical research argue that openness (trade or financial) will definitely increase output growth while others opined that the relationship between the two is ambiguous. In order to contribute empirically to this argument, this study will employ econometric method as the research technique. The choice of method is necessitated by the nature of the study which in this case is an analysis of relationship among variables.
3.2 Model Specification
An economic model is a representation of the basic features of an economic phenomenon; it is an abstraction of the real world (Fonta, Ichoku and Anumudu, 2003). The specification of a model is based the available information relevant to the study in question. That is to say, the formulation of an economic model is dependent on available information on the study as embedded in standard economic theory and other major empirical works, or else, the model would be atheoretical. Two models are postulated in this research work; the first is a non-monotonic model to capture the first and second objectives of the study, while the second is an Analysis Of Covariance (ANCOVA) model. The functional form of the models can be specified as follows:
MODEL I:
RGDPt = f (TPNt, TPNt2, RERt, RIRt, UNEMPt, TREND) … (i)
MODEL II:
RGDPt = f [DUMt, TREND, (DUMt * TREND)] … (ii)
The mathematical form of the models can be expressed as:
MODEL I:
RGDPt = αo + α1TPNt + α2TPNt2 + α3RERt + α4RIRt + α5UNEMPt + α6TREND … (iii)
MODEL II:
RGDPt = β0 + β1DUMt + β2TREND + β3 (DUMt * TREND) … (iv)
But equations (iii) and (iv) above are exact or deterministic in nature. In order to allow for the inexact relationship among the variables as in the case of most economic variables, the stochastic error term "µt" is added to both equations. Thus, we can express he econometric form of the models as:
MODEL I:
RGDPt = α0 + α1TPNt + α2TPNt2 + α3RERt + α4RIRt + α5UNEMPt + α6TREND + µ1t … (v)
MODEL II:
RGDPt = β0 + β1DUMt + β2TREND + β3 (DUMt * TREND) + µ2t … (vi)
Where RGDP = Real Gross Domestic Product which is a proxy for the real output of the economy.
TPN = the degree of openness measured as trade-GDP ratio i.e. (IMPORT+EXPORT)/GDP
TPN2 = the squared term of the degree of openness
RER = Real Exchange Rate
RIR = Real Interest Rate
UNEMP = Unemployment Rate
DUM = 0 for pre-SAP period observations
1 for post-SAP period observations
TREND = the chronological arrangement of time
µ = the stochastic error term.
In order to properly estimate the parameters of the postulated models, we rescale the dependent variable by logging it, thus, transforming them into a log-lin models as follow:
MODEL I:
LOG(RGDPt) = α0 + α1TPNt + α2TPNt2 + α3RERt + α4RIRt + α5UNEMPt + α6TREND + µ1t … (vii)
MODEL II:
LOG(RGDPt) = β0 + β1DUMt + β2TREND + β3 (DUMt*TREND) + µ2t …(viii)
Also, in order to avoid a spurious repression, we subject each of the variables used to unit root (or stationarity) test so as to determine their orders of integration, since unit root problem is a common feature of most time-series data.
3.2.1 Test Of Stationarity
A stochastic process is said to be stationary if its mean and variance are constant overtime and the value of the auto-covariance between the two time periods depends only on the distance or lag between the two time periods and not the actual time at which the covariance is computed (Gujarati, 2003). In other word, a stationary stochastic process is one with constant mean, variance and covariance. Hence, stationarity test is carried out to verify whether a time-series is stationary or time-invariant so as to avoid a spurious regression.
The Phillips-Perron (PP) unit root test will be employed. The choice of this test is to correct for some anomalies associated with the conventional Augmented Dickey-Fuller (ADF) test. The Phillips-Perron test use non-parametric statistical methods to take care of the serial correlation in the error terms without adding lagged difference terms. This test is specified thus:
ΔYt = δ + αΔYt-1 + µt
Where µt is a pure white noise.
Under the null hypothesis that α = 1 for stationarity, we use the PP test statistic to verify the presence of unit root in the series.
3.2.2 Test Of Cointegration
Economically, two (or more) variables will be cointegrated if they have a long-term, or equilibrium, relationship between (or among) them (Gujarati, 2003). Individual time-series in a model may be spurious but their linear combination may not. This is the purpose of co-integration test.
The augmented Engle-Granger (AEG) test will be employed to validate this hypothesis. This hypothesis is of two stages:
We will run the regression of equation (vii) and generate the residual
The residual generated to unit root test.
^
p
^
^If the generated residual is stationary at level form or integrated of order zero i.e. I(0), then the variables of the model are co-integrated. The AEG test is specified as:
i=1 Δµt = δµt-1 + αiΣ Δµt-1 + â„“t
Where â„“t is pure white noise error.
If δ is statistically significant at the chosen level, then the variables of model II in equation (vii) are cointegrated.
3.2.3 Error Correction Model
An important issue in econometrics is the need to integrate short run dynamics with long run equilibrium. The analysis of short run dynamics is often done by first eliminating trends in variables, usually by differencing. But this differencing procedure, however, throws away potential valuable information about long run relationships which economic theories have a lot to say about. The Error Correction Model (ECM) is an extension of short run disequilibrium model, which also incorporates past period's disequilibrium.
The Granger representation theorem states that if two (or more) variables, Y and X(s), are cointegrated, the relationship between (or among) them can be expressed as error correction mechanism. The ECM requires that each time-series is included into the model at its order of integration. Also added to the ECM is the one period lag of the error term generated from the cointegrating regression whose variables are integrated of the same order. The generated error term is treated as the "equilibrium error" and can be used to link the short run dynamics with the long run equilibrium.
The ECM for model I of equation (vii) is specified as:
ΔkoLOG(RGDPt) = Ø0 + Ø1Δk1TPNt +Ø2Δk2TPNt2 + Ø3Δk3RERt + Ø4Δk4RIRt + Ø5Δk5UNEMPt + ECMt-1 + εt …(ix)
The variables are cointegrated if and only if ko = k1, k2, k3, k4, k5
Where Δ = difference operator
ki = order of integration of a particular series
ECMt-1 = error correction mechanism which is the past period equilibrium error.
ε = pure white noise error.
3.3 Justification Of The Model
The choice of a non-monotonic model in this work is triggered by the fact that, though, trade openness at the early stage of introduction into a developing country like Nigeria would certain have a different structure and pattern when compared with its long run effect on the economy's output performance. This fact is embedded in the standard development economic theory and buttress by Baliamoune - Lutz and Ndikumana (2007). The explanation follows suit: at the early stage, when a developing country like Nigeria open up its economy for trade, its domestics firms will face intense competition with the "tiger" foreign firms as the entire market will be flooded with imported products which are cheaper and relatively better in terms of quality than the domestically produced products. This will make some of the infant industries to loss their sales with less revenue along side with high cost of production. This is unlike the foreign firms that enjoy low cost of production with economies of large scale production. As a result, many domestic firms will be forced out of the market. This will surely have a negative effect on the economy.
Nevertheless, as the economy is acquiring new technologies from abroad via openness as well as improving on its domestic infrastructure and capacity utilization of resources in the long run, this will lead to low cost of production for the domestic infant industries and enable them to compete favourably with the foreign products in the market. This will certainly have a positive effect on the economy as its domestic production capacity will increase which will further lead to increase in export products, thus, having favourable balance of payment. Another argument also suggests that developing countries should look inward to achieve its development in the long run.
From the on going discussion, it is evident that fitting a monotonic model for such a situation would either overestimate or underestimate the actual potential of the economy; hence the need for a non-monotonic model. The postulated model is a real model as its variables are all in real form except unemployment. While the degree of openness and real exchange rate represent the external shocks to the economy, the real interest rate and unemployment rate represent the internal shocks to the economy.
Meanwhile, the second model is constructed to test for structural change in the growth of output before and after the introduction of the Structural Adjustment Program (SAP) in 1986.
Finally, the Error Correction Model (ECM) is postulated so as to capture the linkage between the short run dynamics of the economy and the long run equilibrium of the economy.
3.4 Estimation Techniques
In order to develop strong, robust and reliable models that will capture the relationship between trade openness and output growth, the research work adopts the econometric techniques of the NON-MONOTONIC modeling and the Analysis of Covariance (ANCOVA) modeling. In building these models, the Ordinary Least Square (OLS) is used as the estimation technique. The method of OLS is extensively used in regression analysis primarily because it is intuitively appealing and mathematically much simpler than any other econometric technique (Gujarati, 2003). The OLS method is based on some assumptions (see Gujarati, 2003) which make the OLS estimators to become BLUE (Best Linear Unbiased Estimator). Some of the shortcomings of the OLS method include the fact that while some of its assumptions are unrealistic (e.g. No autocorrelation, homoscedasticity and No multicollinearity), a single model as well can not fully satisfy all the assumptions at a time. Also, no single test can solve all the problems of this method at a time. Moreover, the OLS method can not be applied to intrinsically non-linear model such as one that is non-linear in parameter. As a result of some of these short-comings, we use the OLS method but correct the standard errors for autocorrelation by a Newey - West method. The corrected standard errors are known as HAC (Heteroscedasticity-and autocorrelation-consistent) standard errors or simply as Newey-West standard errors.
Hence, we have to apply individual initiative along side with the empirical rules and tests so as to obtain tenable and robust results. Thus, an econometric modeling is said to be more of an art than a science.
3.5 Evaluation Procedure
3.5.1 Economic Test (A Priori Expectation)
Tests shall be conducted to ascertain the a priori expectations which examine the magnitude and signs of the parameter estimates. This evaluation is guided by economic theory. The aim of this test is to confirm whether the parameter estimates conform to a priori expectation. The variables used in the model and their a priori expectations are analyzed below in table (2).
Variables
Definition
Expected sign
RGDPt
This is Real Gross Domestic Product which represent the real output of the economy. It's natural logarithm is taken.
It is the dependent variable and considered to be stochastic.
TPNt
This is the degree of trade openness in an economy. It measures the international competitiveness of an economy in the global market. It is an external shock to the economy.
It is expected to be positive.
TPNt2
This is the squared term of the degree of trade openness. It shows the structural pattern of openness relative to output growth.
Since the structural pattern could be of any type, it can be positive or negative.
RERt
This is the real exchange rate. It is the rate at which the domestic currency is being exchanged for the foreign currency with adjustment for relative price index. It is an external shock to the economy.
It is expected to be negative.
RIRt
This is the real interest rate. It is the real cost of borrowing fund in the financial market. It is an internal shock.
It is expected to be negative.
UNEMPt
This is the unemployment rate in the economy. It is an internal shock.
It is expected to be negative.
DUMt
This is a dummy variable introduced to capture the effect of the Structural Adjustment Programme (SAP) trade deregulation and liberalization policies.
Since the effect of a policy could be favourable or adverse, its sign can be positive or negative.
TREND
This is the chronological arrangement of time which captures the incremental growth of output overtime. It also serves the purpose of detrending the fluctuations among the exogenous variables.
It is expected to be positive.
ECMt-1
This is the error correction mechanism
It is expected to be negative.
Table : A priori expectation
3.5.2 Statistical (First Order) Test
Here, various statistical tests will be carried out so as to verify the acceptability, reliability and robustness of the estimated regression result. The tests include:
Student t-Test
This is used to test the statistical significance of the individual parameter estimates in the regression models. This work will use the t-distribution to test the statistical significance of these parameter estimates.
F-Test
This is used to test for the overall significance of the model. It tests the simultaneous null hypothesis of all the parameter to be equal to zero in the regression model.
R2 - Coefficient Of Determination
This test is used to measure the goodness of fit of a regression line. It measures the proportion of the total variation in the dependent variable explain by the regressors in the model.
r - Coefficient Of Correlation
This measures the significance of the strength of linear association (correlation). The correlation coefficient measures the degree of association between two variables (such as TPN and RGDP in the case of this work). It is obtained through the Product Moment Correlation Coefficient.
3.5.3 Econometric (Second Order) Test
Here, various tests will be carried out in order verify whether the estimated regression results conform to the Classical (Normal) Linear Regression Model assumptions. This test includes:
Test Of Normality
This test is used to verify whether the error term is normally distributed. The Jacque-Bera (JB) test will be used to verify this assumption.
Test Of Heteroscedasticity
This test is used to verify the assumption of equal spread of the error variance (homoscedastic) between members of the same series of observations. The White's heteroscedasticity test (with no cross term) will be employed in the test.
Test Of Autocorrelation
This test is used to verify the randomness of the error term between members of the same series of observations. As a result of the numerous assumptions and problems associated with the conventional Durbin-Watson (DW) test, the Breusch-Godfrey (LM) test will be employed to verify this hypothesis.
Test Of Specification Errors
This test is used to verify whether the econometric regression model being estimated is correctly specified. The Ramsey's RESET (Regression Specification Error Test) will be employed.
Forecast Test
This test is used to verify the reliability of the estimated regression model in forecasting future values. The Henry Theil's inequality coefficient will be used to evaluate the forecasting performance of the model.
3.6 Source Of Data
Data is the most important material for any economic research or analysis, and very much indispensable to the field of econometrics indeed. Gujarati (2003) asserted that the success of any econometric analysis ultimately depends on the availability of appropriate and accurate data. In other words, the researcher should always keep in mind that the results of research are only as good as the quality of the data.
The research study makes use of secondary data. The data used are obtained from the Central Bank of Nigeria (CBN) statistical bulletin 2007 and National Bureau of Statistics (NBS) online publication for 2006.
The percentage ratio and real values are computed by the researcher in order to capture the objective of the study and in congruence with economic theory.
3.7 Software Package
The econometric software package used for the analysis of this work is the Eviews3.1 version while the Microsoft Excel 2003 is used to enter the data.
3.8 Chapter Summary And Prospects
In this chapter, we have discussed the methodology to be employed in carrying out the analysis of the work. We discussed the model specification, the justification for such model, the estimation techniques to be used and the evaluation procedure which includes various economic, statistical and econometric tests. We concluded the chapter with the sources of data and the software packages to be employed for analysis of data. In next chapter, we will present the regression results and evaluate the results using the aforementioned tests.
