Testing the Productivity Growth Effects of Financial Liberalisation on the Economic Sectors of Ghana in the Precence of Structural Breaks

The study examined the relationship between financial liberalization and productivity growth of the three economic sectors in Ghana by using the yearly data over the period 1970-2013. In the analysis; the credits provided to private sector, investment, trade liberalization and capital account openness are considered as indicators of financial liberalization while sector level value added as a percentage of GDP represents productivity growth. The stationarity of the series and the long run relationship are analyzed by means of Zivot-Andrews (1992) and Clemente, Montanes and Reyes (1998) tests and Gregory Hansen tests in which structural breaks are considered. The findings of the study revealed that increasing spending on the productive sectors of the economy will yield a positive result of sustainable productivity growth at sector levels. However, government initiated structural reform that is not accompanied by sound macroeconomic policies may not have a lasting effect, since the effect of such reform in the long run growth path could be thrown out of gear by other external shocks. The positive effect of financial liberalization on productivity growth can be broadly confirmed in all the three economic sectors. Future studies could be focused on whether financial liberalization will make anticipated impact on sector level productivity growth using primary data from the various sectors of the economy in a survey study using structural modeling.


Background
Mounting empirical and theoretical evidence demonstrates that cross-country differences in income disparity and economic development can be explained by differences in the productive capacity of firms and economic sectors of the countries in the global economy and particularly differences in productivity between developed and developing countries. Standard economic theories on the issue from various authors have made attempt to offer insight on income disparities employing factor accumulation and total factor productivity (TFP). Both Single-country and cross-country studies in development accounting has also demonstrated how TFP explains income per capita. Available literature shows that TFP accounts for 50% -70% of per capita income differences across countries (Hsieh & Klenow, 2010). Foster et al. (2001) also contend that reallocation among firms contributes immensely to aggregate productivity growth. As a result the low productivity observed in developing countries could be explained from the point of view of developing countries inability to allocate production factors from less productive to more productive firms and sectors (Bartelsman et al., 2013). Economic literature provides fresh insight into how cross-country capital flow and progressive development of domestic financial activities influences generation of savings for investment activities (McKinnon, 1973;Shaw, 1973) which in turn contribute to more efficient allocation of scarce economic resources (Bencivenga & Smith, 1991;De Gregorio & Guidotti, 1995;Goldsmith, 1969). In addition to this, well developed financial market transactions promote efficient risk diversification between alternatives economic purposes.
The argument presented here has in most cases formed the basis for politically-driven interventions to embark on financial liberalization process worldwide, both involving developed and developing countries. Empirical and theoretical literature on the nexus between financial liberalization and growth indicates that financial liberalization on average stimulates productivity growth. Bekaert et al. (2005) in their study provided evidence to confirm this positive link both when treating the growth process in general terms (Bekaert et al., 2005) and also in the case concerning decomposition of the impact on the two main sources, productivity and capital accumulation (Gehringer, 2013).
In the 1970s, the phenomenon of financial repression, a major economic policy that was adopted by many governments especially in developing economies for the purpose of generating growth and revenue by means of artificially low interest rates and inflationary monetary policies attracted stronger discussion on the issue. The argument had its foundation from the theoretical works of Keynes (1936) and Tobin (1965), who proposed the need for government interference in the credit market. However, evidence from both single-country and cross-country studies reveals that a range of different financial frictions retard productivity growth by obstructing optimal allocation of economic resources. Empirical evidence from other studies (Levine &Warusawtharana, 2017;Buera et al., 2011;Hall & Jones, 1999) also suggests that financial frictions can explain a sizeable portion of the differences in productivity between developed and developing countries. Therefore, the argument that follows is that, the opening of financial services coupled with structural reforms of domestic financial sector constitutes the essential processes intended at developing an efficient and competitive financial system towards enhanced productivity growth. Thus a good financial system running hand in hand with well-functioning competitive market together with a well-supporting financial institution provides an enabling environment for sustainable productivity growth. Schumpeter (1911), buttressing this point contended that the services provided by financial intermediaries are among the essential ingredients required by both developed and developing countries for driving innovation and growth. This argument was formalized by Mckinnon (1973) and Shaw (1973), and supported by their contemporaries Fry (1988) and Pagano (1993). It was Mckinnon (1973) and Shaw (1973) who posited that government restrictions on the operation of the financial system in terms of interest rate ceiling, direct programs and high reserve requirements has the tendency to impede financial deepening. This they contend may negatively influence the quality and quantity of investments activities and hence retard productivity growth. With regard to literature on the nexus between financial liberalization and growth, some authors believe that the relationship has been "badly over-stressed" (Lucas, 1988, p. 6). Arestis and Dematriades, (1999), considering the analysis of the issue with regard to developing countries argue that unrealistic assumption of perfect information and perfect competition could be the main cause of the discrepancy between financial liberalization theory and evidence.
Ghana, as in the case of several other developing countries has implemented various structural and economic reforms programmes aimed at reaping the benefit associated with well-developed, effective and efficient financial system. These structural reform programmes were introduced at the time when the country's financial sector was faced with financial repression leading to the abysmal performance of the sector as an intermediary with the role to facilitate the attainment of growth levels in real sectors (manufacturing and agriculture) of the economy (Adu et al., 2013). The Financial Sector Adjustment Program (FINSAP) was then introduced in 1988 to help liberalize the financial sector which was facing several challenges in the early 1980s. The results were significant and positive effect on various financial systems which lead to the growth of the economy over the years of implementation. This study is intended to investigate the influence of financial liberalization on sectoral growth performance of the main economic sectors of the economy. Specifically, the study will distinguish between the productivity growths as measured by TFP of the three main economic sectors of the country. It further examined the nature of structural breaks and external shocks to financial liberalization and productivity growth of the three economic sectors.

Problem Statement
The Ghanian economy, like all the Sub-Saharan African (SSA) countries was adversely affected by the international recessions, debt crises and political instability that severely hit the sub region in the 1970s. The economic condition at the time which was occasioned by a reduction in the domestic resource mobilization and weak tax base, further depressed investment and economic growth leading to a deterioration in economic performance in the presence of financial repression during the period which gave rise to the adoption of structural adjustment programme. In Ghana, the economic conditions resulted to financial sector reforms as part of the Structural Adjustment Programme which was embarked upon in 1983 and 1996. Successive governments in Ghana over the years saw it as a matter of urgency to initiate and implement series of financial reform measures in an attempt to stimulate sustainable growth and economic development. The motivation for the implementation of these reform programmes was on the basis of a plethora of worldwide studies showing evidence of the dynamics of international financial liberalization giving impetus to rapid aggregate growth in support of the debate which favors a positive relationship between financial liberalisation and improved economic performance (Gehringe, 2013). Depending on the theoretical approach that one applies, financial liberalization is deemed to have positive or negative effects on the growth process. This assertion is confirmed both when analyzing the growth process in general terms as in Bekaert et al. (2005) and when the intention is to investigate the impact on its main sources, productivity and capital accumulation (Bonfiglioli, 2008;Gehringer, 2013).
Contrasting the view expressed above with available literature (Demirguc-Kunt & Detragiache, 1998;Kaminsky & Reinhardt, 1999;Stiglitz, 2000;Glick & Hutchinson, 2001) which concerns more specifically with the economics of financial crises provides ample reasons to believe that financial liberalization might be the main cause of increased frequency of crisis and of induced macroeconomic volatility. The aforementioned analyses emphasized exclusively an aggregate perspective, putting together a wide range of economic sectors, industries and firms in various developed and developing economies. Moreover, macro data used in the empirical investigations were considered adequate to grasp an average effect of the phenomenon. Only a few studies recognized the need to make use of more disaggregated view: either by applying sector-level data (Levchenko, 2009;Becker et al., 2011) or by investigating a particular country, like Ghana being the case in Djokoto (2013), Karikari (1992), Gyapong andKarikari (1999), Frimpong andOteng-Abayie (2008), and Sakyi (2011) who documented empirical evidence for the whole economy of Ghana on the relationship between financial liberalisation and economic performance, and Adenutsi (2008) for industrial performance also in Ghana. Yet there is still an unexploited area of the story that needs to be addressed.
In Ghana, studies that attempt to investigate the impact of financial liberalization on productivity growth which is considered as one of the main source of economic growth are not readily available (Djokoto, 2013). More precisely, studies that make effort to investigate and quantify the effect of financial liberalisation on the productivity growth distinguishing between manufacturing, service and agriculture sectors are completely absent from Ghana, making this research the first of its kind in Ghana. Furthermore, Gehringer (2014) found that due to the diversity in the intrinsic characteristics of manufacturing, service and agriculture sectors, their individual sector responses to the liberalisation processes are also expected to differ. This is because the impact of financial liberalisation is found to be sector-specific or at least differ in its effects on manufacturing, service and agriculture sectors (Gehringer, 2014).
This study is intended to investigate the effects of financial liberalisationon on productivity growth measured by total factor productivity (TFP). The study, analyses manufacturing, service and agriculture sectors in separate with the aim of preventing any sector-specific characteristics from influencing the determination of the more general relationship between financial liberalisation and productivity growth. Moreover, the study investigate the influence of financial liberalisation on the sectoral productivity growth particularly, distinguishing between the effects of financial liberalization on the productivity growth of the three main economic sectors (manufacturing, service and agriculture) of Ghana. (1956) and Swan (1956). According to the neoclassical theory, technological progress is determined by scientific processes that are exogenous to economic forces that bring about economic growth. This assertion is challenged by the endogenous growth theory which specifies the channels through long-run economic growth fueled by technological progress that can be influenced by economic factors (Kaldor, 1957;Koushik, 2017).

Financial Liberalisation Theory
The seminal work on finance-growth nexus presented by Schumpeter (1911) was subsequently followed by several other studies such as that of goldsmith (1969), McKinnon (1973) and Shaw (1973). These studies have formed the bedrock of the theory analyzing the effects of financial liberalization on productivity growth both in developed and developing economies. The studies by Schumpeter (1911) which is generally referred to as the initial framework leading to the analysis of the link between finance and growth hypothesis maintained that having a well-functioning financial system in place is an important first step in stimulating technological innovation (growth) through efficient allocation of economic resources flowing from unproductive to productive sectors within an economy. Schumpeter's argument is supported by Patrick (1966) supply-leading hypothesis which also maintained that when well-functioning financial market and associated services are created well in advance of the time they are needed, they draw along with it the real sectors on the path of growth, by means of transferring scarce resources from surplus spending units to deficit spending units in accordance with the highest rates of return on investment.

Measurement of Productivity
Productivity can broadly be described as the efficiency with which inputs are turned into output (Hulten, 2001). Productivity has been considered as a derived measure, and therefore not a straightforward concept that can directly be observed. Owing to this reasons, its calculation can be approached from different angles. The method of calculation chosen much depends on the purpose for which it is measured, the unit of measurement applied and availability of data (OECD, 2001).
Productivity can be measured using single factor productivity measure; example is labour productivity (OECD, 2001). With this approach output measurement is usually related to a single measure of input. Alternatively, productivity can also be measured using multifactor productivity measure as with Total Factor Productivity (TFP). Here a measure of output is related to observable group of input. In each of the two measures outlined above, there is also an issue of the choice of inputs and outputs; where an output is measured by means of gross measure or value added approach (OECD, 2001).

Total Factor Productivity (TFP)
Total factor productivity (TFP) is acclaimed to be the measure that presents comprehensive measurement of productivity. TFP, according to Kathuria et al. (2011) is defined as the ratio of output (or value added) to a weighted sum of the inputs used in the production process. Due to the nature of its inclusiveness, it is not confronted with the problem of changing factor intensities in production. TFP measurement takes into account both intermediate inputs and capital intensity.
Even though TFP has been subjected to criticism as a concept for measuring firm or industry level productivity, literature abound with several authors who employed it as a measure of firm level or industry level productivity (Gatti & Love, 2008;Grilli & Murtinu, 2011;Colombo et al., 2011;Harris &Li, 2012;Ackah et al., 2012;Levine & Warusawitharana, 2014).

Measurement of Production Function
The issue regarding the estimation of production function and productivity is the question of how input and output should be measured. There are however, several alternatives available. Authours such as (Saliola & Seker, 2011;Ahiakpor & Dasmani, 2012) approached the issue from gross output measures using revenue or sales. Value added approach was employed by others (Ospina & Schiffbauer, 2010;Levine & Warusawitharana, 2014). Besides these, (Biatour et al., 2011;Ackah et al., 2012) chose both gross output and value added approaches.
None of the three approaches described have been rated superior or preferred to the other. Whichever approach is used will depend on the availability of data. The value added approach can be conveniently used if data on intermediate inputs is available. To be able to use sales or revenues approach what one requires is to deflate their values using price deflator series where a given year is considered as the base year (Ackah et al., 2012).

Financial Liberalisation and Productivity Growth
The functions performed by financial system in any economy in developed or developing countries cannot be overemphasized. A hand full of the important functions performed by financial system are given below: Among the basic functions are mobilization of financial resources for productive activities, financial system facilitate risk management, resource allocation from less productive to the most productive activities in any economy is imperative in enhancing productivity growth. This is an aspect of intermediation process that is facilitated by financial system (Levine, 1997).
The theoretical foundation of the argument supporting the assertion that well developed financial system has a greater tendency to enhance productivity growth in both developed and developing countries was laid by Schumpeter (1912). The views expressed by Schumpeter (1912) have been confirmed and enhanced by subsequent studies (Greenwood & Jovanovic, 1990;King & Levine, 1993). The import of the argument developed by Schumpeter (1912) is that, significant among the beneficial consequences of financial liberalization is its ability to spur allocative efficiency of savings which eventually lead to productivity growth. This is generally achieved through information channel, whereby well-developed financial intermediation process minimises the cost of information to the extent that savings are brought to where they are mostly needed for competitive productive activities (Beck et al., 2000). It is argued that where better information are made available to the financial intermediaries concerning the destination of productive and potential innovators, credit constraint may not hinder production activities which are even classified as high risk investment. This eventually will lead to advancement in technological progress (Acemoglu et al., 2006;Acemoglu & Zilibotti, 1997) and eventually productivity growth.
Apart from economic policies that deal with the elimination of financial distortions, other studies Aghion et al., 2005) have revealed that capital market development has a closer link with higher productivity growth. Reforms that encourage the formation and development of equity, bonds and securities markets can be effective in increasing productivity by means of bringing down cost of capital and making funds available for new capital and innovation. In other related studies, it was discovered that availability of financial instruments towards financing innovation process are mostly beneficial to countries closer to the technology frontier (Aghion et al., 2005).
Evidence found by  demonstrates that structural reforms enhance resource allocation leading to a boost in productive capacity. They discovered that productivity growth in EMDEs is dependent on catch-up growth by absorbing technology and ideas from technology leaders, structural reforms toward higher-productivity sectors and efficient resource allocation within sectors. A large body of other empirical evidence in a related study abound in literature on how trade openness, efficient and well-developed financial systems, and economic institutions with the capacity to promote competition and facilitate entry and exit, increase productivity growth at the cross-country, sector, and firm levels (Christiansen et al., 2013;OECD, 2013;Prati et al., 2013;Restuccia & Rogerson, 2013;Bourles et al., 2013).

Emperical Literature
Empirical studies on the effects of financial liberalization on productivity growth have yielded different results; various factors have accounted for this result. Among other factors, the following can be considered: estimation methodology (Krishnan et al., 2014, Nandy & Puri, 2014, stage of development of the economy (Chauvet & Jacolin, 2015;Gatti & Love, 2008), data used and sample size (Ali & Najman, 2015).
Empirically, the nexus between financial liberalization and productivity growth has been confirmed by several researchers on the subject. Beck, et al., (2000) examined the nexus between financial intermediary development and sources of growth. They used private credit and liquid liabilities as proxies for measuring financial intermediary development for 63 countries for the period 1960-1995; they discovered a strong positive link between financial intermediary development and real per capita GDP growth and total factor productivity growth.
Evidences drawn from both theoretical and empirical studies reveals that efficient financial systems cause increase in investment and promote innovation (Levine, 2005). In this sense, countries running efficient financial systems are therefore able to take advantage of technology transfer. Efficient capital allocation across firms and industries follow well developed financial systems (Rajan & Zingales, 2001;Tressel, 2008). In countries where financial repression and restriction on price or quantity of credit have been suppressed, it is known that resources move freely from across and within productive sectors. Larrain and Stumpner (2013) identified this evidence in 10 Eastern European countries that reforms directed at reducing financial repression have been found to increase manufacturing productivity by 17% resulting from enhanced allocation of resources across firms and within-industries.
Cross-country analysis offers more insight into the extent to which productivity payoffs of financial liberalization may vary across countries at different developmental stages (Chipeta, 2012). However, access to credit and its effects on productivity growth was not found to differ across countries at different ijbm.ccsenet.org International Journal of Business and Management Vol. 15, No. 4;2020 development levels. Specific reference was made to some industrialized countries such as Bulgaria, China and other developing countries in Africa such as Tanzanian. Evidence provided by Krishnan, Nandy and Puri (2014) shows that where smaller and financially constrained firms are given access to bank financing, their productivity is affected positively. Contrary to other studies (Krishnan et al., 2014;Chauvert & Jacolin, 2015), theirs was based on natural experiment of US private and public manufacturing firms, subsequent to the interstate banking deregulation which immensely opened bank financing to firms. Their findings was not different from that of Robb and Robinson (2014) who also found increased access to bank finance as a very critical factor in firms' productivity, most especially with regard to small and startup firms. It is also discovered that line of credit and savings has a strong positive link with productivity growth measured by Total factor Productivity (TFP) (Ali & Najman, 2015).
Studies in African and other developing countries have not yielded contrary evidence. Ali and Najman (2015) observed in a large survey of 33 African and other developing countries and discovered that where the perceived access to finance remains an obstacle for firms in those countries the less productive they become. On the other hand the same firms when given more access to external finance become better off in terms of their productivity. This view has been supported by International Finance Corporation(IFC) (2013) which also identified financial constrain as a major obstacle to enterprise growth and productivity among small and medium sized firms.
Osei-Assibey (2013) also found that in Ghana, Medium and small enterprise having increased access to debt finance or loan achieve higher productivity in their operation. His study was based on data sourced from the Ghana Living Standards Survey (GLSS 5).. Chauvet and Jacolin (2015) rather distinguished between firm level and country level determinants in assessing the factors that influence both firm level and country level determinants of productivity. They identified that statistically, foreign ownership, size, export and overdraft facility significantly serve as a strong firm-level determinants of productivity, however GDP growth, income per capita, population and corruption ware considered as country level determinants.
In a related study by Andrews and Cingano (2014) it was found that well developed financial system moving hand in hand with lesser stringency in banking regulation leads to higher productivity. According to the authors, this is achieved by means of net firm entry mechanism. The implication is that availability of finance enhances productivity growth through the elimination of lower productivity firms in the economy. Their results agree with findings of similar study by Midrigan and Xu (2014). Bravo-Biosca et al. (2013) also found an indirect evidence of link between firm-dynamic-related productivity and greater financial development in a study carried on in 10 OECD countries using firm level data covering a period of 2002 to 2005. The conclusion drawn was that in the case where industries depend much on external finance, well developed financial institutions within an economy yields a more dynamic growth distribution of firms, a condition that promotes reallocation of capital through the firm churning process.
In a study conducted by Cole et al. (2016) in which firm technology adoption decisions for three countries including India, Mexico and the United States were modeled using external financing as a critical determinant, it was revealed that a financial system that engages in long-term contracts and efficient performance monitoring is a key driver for the adoption of advanced technologies that promotes higher productivity. Advanced production technologies require higher initial capital outlay with the payoff late in the firm development cycle. Countries such as US having efficient financing institutions and mechanisms are able to take advantage of adopting such advanced technologies. In the case of India and Mexico running flexible systems which permit adoption of entry-level and intermediate-level technologies, their financing needs are lower with shorter pay-off horizons as compared to advanced technology. The basis of differences in TFP between countries can therefore be highlighted by the kind of technology choice adopted. They further hinted that TFP for India and Mexico have the potential to rise by 46% and 43% respectively if their financial system were as well-developed as that of the United States (Cole et al., 2016;Ali & Najman, 2015).
Adusei (2013) on the other hand applied annual time series data from 1971 to 2010 in carrying out a related study in Ghana. Unlike the previous study, he used The Fully Modified Ordinary Least Square (FM-OLS), Error Correction Model (ECM) and Generalized Method of Moments (GMM) for estimations. The study found that finance impedes economic growth in Ghana. His conclusion was that financial liberalization undermines growth and therefore not beneficial to the economy. The study further revealed that FM-OLS and ECM results confirms that money supply to GDP ratio and domestic credit to GDP ratio hinder growth in both the long-and short-run (Adusei, (2013)

Research Design
The research design for this study is quantitative design. The study is intended to conduct a systematic and objective examination of the variables involved in the study in order to establish and measure the relationship between financial liberalization and productivity growth, using sector-level data. The study used multivariate analysis model.

Statistical Data Analysis/Procedures
The analysis of data is based on econometric models such as Error correction model to examine the short run and long run relationship among the variables; Multivariate regression analysis, which involves the study of the relationship among the variables in the model, will be applied (Lancaster, 2005).

Econometric Estimation Techniques
The investigation and examination of the long run relationship is based on testing for unit root; performance of cointegration test (Gregory and Hansen model).

Nature and Sources of Data
The data used in this study are secondary data. Only quantitative data was used in this study

Measurement of Productivity
Several forms of Production function are available for estimating Total Factor Productivity (TFP). Among others, the following can be considered: Transcendental Logarithmic (Translog) Production Function, Constant Elasticity of Substitution (CES) and Cobb-Douglas production function. Following other authors such as (Chen & Guariglia, 2011;Osei-Assibey, 2013;Gatti & Love, 2008;Arnold et al., 2008). This study employs the Cobb-Douglas Production Function in Estimating Total Factor Productivity (TFP).

Estimating Total Factor Productivity (TFP)
This study used the Value-Added-Based Total Factor Productivity measure.
Estimation of total factor productivity measure begins with the standard Cobb-Douglas production function given as equation (1): (1) Where TFP it denotes technology contribution to output (Y) in s e c t o r i a t time t, β coefficients refer to each factor's (X) estimated output elasticity and m= (1, 2…m) Equation (1) is then transformed into a log-log model as in equation (2) The productivity level is obtained as the residual of the (lnTFP it ) by subtracting from output the observable inputs contributions weighted by their corresponding output elasticities Going by the value added approach and given that value added is defined as the difference between the value of gross output and the value generated by intermediates (X 1 …X M ): The log productivity measure for each Sector-year observation, TFP st is taken as the residual from analogous log-log model as in equation (3): International Journal of Business and Management Vol. 15, No. 4;2020 In this sense, TFP growth (lnTFP it -lnTFP it-1 ) is defined as the value added not influenced by an increase in input (labour and capital) The regression estimation below is therefore based on equation (4) (t), and K st and L st denoting capital and labor inputs respectively. The log specification given in equation (3.1) assumes a Cobb-Douglas model for value-added by economic sectors (Manufacturing, service and Agriculture). The intention here is not to impose a constant returns-to-scale restriction in the above model, however, both the capital and labor elasticities will be estimated.
The log productivity measure for each Sector-year observation, TFP st is taken as the residual from equation (3).
The Log growth rate of Productivity which will be employed as the dependent variable for the subsequent models is derived by taking differences of this residual. This is expressed as: In the model, capital is measured using the replacement value of capital while labor measured as the staff cost of employees. The replacement value of capital is computed based on double declining balance method used by Summers and Salinger (1983). This approach (replacement value of capital) allows the incorporation of potential differences in the quality of the capital stock. Using staff cost of labour in the measurement of labour input also permits the author to capture potential differences in the skill level of employees The TFP measure that is estimated indicates the amount of value-added by various sectors beyond what can be attributed to its capital and labor inputs. Therefore, an increase in TFP will mean that where input factors are held constant, the economic sector involved generates a higher amount of value-added to productivity.

Model Specificatio
The next step after estimating productivity (TFP) is to regress on its policy variables. The main policy variables to be considered in this study are financial liberalisation proxied by Credit to Private-Sector, Sector-Level Investment (FDI) and Trade openness. There are other potential determinants of sector level productivity growth identified in literature such as Capital Account openness, Market Capitalization and Credit Control Relaxation). That notwithstanding, empirical theory does not indicate any particular determinant which must be used in an empirical model. The study will therefore follow the stepwise approach whereby as many significant variables as the data may provide can be included in the model.
The baseline model for this study is based on the endogenous growth model. Following Adusei (2013) and also on the basis of the theoretical literature, the model for the study is specified as: = + + + + + With TFPit+1 denoting real productivity growth from year t to t + 1; Finlib st is the financial liberalization index, X st denotes a set of additional controls; and ν s denotes a sector fixed effect. The coefficient of interest, γ, is interpreted as an elasticity term. The lagged productivity growth terms are included for the purpose of accounting for the mean reversion in productivity observed in the data. ε is the idiosyncratic error term. A positive coefficient for γ signifies that financial liberalization promotes productivity growth within the sectors. The control variables used comprise sector's size, physical investment during year t, government expenditure, market capitalization and credit control relaxation. Considering the significance of these control variables, a positive relationship with productivity is expected. Government expenditure in percentage of GDP included as part of the control variables measures the contribution of the government involvement to economic growth. This influence even though, is expected to be positive; it might also be negative due to the crowding out effect regarding private investment (Adusei, 2013).
A sector's size variable is also included for the purpose of soaking up some sector-specific characteristics in relation with a sector in an economic system (Maskus et al., 2012;Friedrich et al., 2012).

Estimation Strategy
For most time series regression analysis, consistency in the data and parameter estimates is very relevant. Therefore, the study follows three major steps to estimate the parameters included in the estimable model. Firstly, the study examines the stationarity properties of the variables included in the model. Secondly, the study tests for the presence or otherwise of cointegration (long-run relationship) amongst the variables. Finally, the study estimates the long-and short-run parameter estimates.
Where γ is time trend, Y is time series variable in the model, ɛ t is the error term or stochastic error term. Given that Y t is the level of the series variable, μ is the drift term, T is the time trend, P is the number of lags, ∆ indicates that the series are in their first difference. The ɛ t is the error term/ white noise having the features of normal distribution. The error term (ɛ t ) has an expected mean value of zero (0) and a constant variance. None of the errors is dependent on each other. The constant term and time trend are not present in the equation (2), equation (3) rather has a constant term while; equation (4) has both the constant term and time trend.
The null hypothesis (H 0 ) states that the series are non-stationary in levels; while the alternative hypothesis (H 1 ) states that the series are stationary. The critical values are compared with the computed values at 5%, 1% and 10% levels of significant.
The ADF Test is governed by the decision rule which states that if the computed ADF value in absolute term is less than the ADF critical value in absolute term, the series in the model are said to be non-stationary or has unit roots, which therefore requires differencing until they become stationery.

The KPSS Test
The KPSS test is normally performed for the purpose of confirming the ADF test. The main null assumption underlying the KPSS test is that the series variables being tested are stationary as against the alternative assumption which states that the series are non-stationary (Kwiatkowski et al., 1992).
The KPSS Test is expressed in an equation form given in equation (10 μ − = + Given that µ t is equal to IID (0, σ2 μ ). r t has an initial value of zero (0) given by r 0 and also considered as fixed. It functions as the intercept in the model. The stationarity assumption governing this test is given as σ2 μ =0. The series variable (X t ) which is being tested assumes trend stationary properties on the basis that the error term is stationary.

Unit Root Test with Structural Breaks
The analysis of time series variables with structural breaks are based on several tests including Zivot-Andrews (1992) and Clemente, Montanes and Reyes (1998) tests. The examination and investigation of unit root with one endogenously determined structural break is examined with the use of The Zivot-Andrews (ZA) unit root test. This test estimates the break date but does not consider it as fixed. The ZA test allows only a single break in the intercept and the trend of the series variable being examined.
The null assumption specified in equation (15b) states that the time series variable being examined is integrated without an exogenous structural break while the alternative hypothesis states that the series is trend-stationary with a break point occurring in the trend at an unknown point. In this sense break point can be estimated at a point that favours the alternative hypothesis. . The t-test is also used in addition to the selection processes outlined above. In the case where the series variables exhibit multiple breaks, the ZA test becomes limited in dealing with it. An appropriate test was developed by Clemente, Montanes, and Reyes (1998) which permits multiple structural breaks to be examined.
The assumptions (Null assumption and the alternative assumption) underlying the test are specified in the equations (17) and (18) μ θ θ ε = + + + Given that DTBit is the dummy variable. DTB it is equal to 1 if t= TB i +1, 0 otherwise. DU it = 1 if t˃TB i (for i=1, 2), 0 otherwise. TB1 and TB2 represent the period for the break point. Clemente et al. (1998) specified that TBi = λiT (for i=1, 2) for the range 0˂λ i ˂1 and λ 2 ˃λ 1 .

The estimation of the break points is made dependent on innovative and additive outliers.
Equation (19) The assumption for the asymptotic distribution of the test statistic for the above model is given in a model (20) below. Based on the assumption that λ 2 ˃λ 1 +1, the issue regarding the case where the breaks occur at conservative period can be eliminated.
Structural break involving additive outliers can be examined using the models specified in equations (21) (22) is then estimated on an assumption that ρ =1. The minimum value of t-ratio is employed in testing for the null assumption that ρ =1.
Where DTB it is the dummy variable included in the model to ensure that min tA0 ρ^ (λ 1 , λ 2 ) converges to the distribution.

Cointegration Analysis Base on the Gregory and Hansen Methodology with Structural Breaks
The Gregory and Hansen Methodology of cointegration permit the investigation of long run and short run relationship with endogenous structural breaks. The four models specified in equations (24, 25, 26 and 27) are specifically developed for this purpose. The investigation is based on the following assumptions: null assumption that there is no cointegration with structural break against the alternative assumption that there is cointegration with structural break.
The variables defined in the equations: M and N are the two series variables with M denoting the dependent variable and N denoting the independent variable. e is the error term t is time subscript; and k is the break date.  (24) to (27), require an additional equation (28) as specified below.

Estimation of dummy variables in equations
f tk = 0 if t ≤ k and f tk = 1 if t ˃ k …………… ……. (28) The determination of the break date depends on the estimation of the cointegration equations for all possible break dates in the series under investigation. The minimum t-statistics is used in selecting the break date or ADF test statistic is maximum in absolute term. Critical values in Engle-Granger model developed by Gregory and Hansen (1996) is employed in investigating cointegration in the presences of unknown structural breaks.

Time Series Plot of the Variables in Levels without Breaks
The Time Series plot results shown in Figure 1 (Appendix A) indicate that, all the series are not stationary in levels, and might achieve stationarity by differencing (Figure 2) (Appendix B). This implies that there is the need for further scientific investigation of the nature of unit root using the KPSS model of unit root.

The Unit Root Tests without Structural Breaks
The series were examined in levels in linear form, and in first difference. The results in Table 2 indicate mixed results. Some series are unit root in levels but become stationary in first difference as indicated in Table 3, with the implication that they are integrated of order one, I (1). Series variables that are stationary at levels are integrated of other zero, I (0). The levels of significance are 1%; 5% and 10%. Some series are stationary at 10% but not at 1% and 5%. The results indicate that all the series are stationary in first difference. The KPSS is a reversed test for unit root. It is used for confirmation of the stationarity properties of the series.

Testing Unit Root Allowing for Endogenous Structural Breaks
The Zivot-Andrews test was used to test for unit root allowing for endogenously determined structural break.
The results are reported in Table 4. The test is based on the null hypothesis of unit root against the alternative hypothesis of no unit root. The null hypothesis of unit root cannot be rejected excerpts Agriculture Sector Value Added (ASP) and Service sector value added (SSP). The series variables are unit root with structural breaks. The break dates coincides with known national dates in the Ghanaian economy.  --4.42) and 10% (--4.11): Note: ***, ** and * denotes statistical significance at the 10%, 5% and 1% levels respectively.  --4.42) and 10% (--4.11): Note: ***, ** and * denotes statistical significance at the 10%, 5% and 1% levels respectively.

Testing for Unit Root Allowing for Two Structural Breaks
The Clemente-Montanes-Reyes test (1998) was used to test for unit root allowing for two structural breaks. The Clemente-Montanes-Reyes test has two models, which are the innovational outlier (IO) and Additive Outlier (AO). The results for IO model are reported in Table 8 and that for AO model are reported in Table 7. According to the IO model, structural changes take place gradually and allows for a break in both the slope and the intercept of the model. In the AO model, the assumption is that structural changes is rapid and allow for a break in only the slope. The test is based on the null hypothesis that the series are not stationary with structural breaks (there is unit root) and the alternative hypothesis that the series are stationary with break (there is no unit root). Despite the structural break in the series using the innovative outlier, Manufacturing Sector Value Added (MSP) and Service sector value added (SSP), we are unable to reject the null hypothesis of a unit root in the series variables in levels, as shown in Table 7. In Table 9 we are able to reject the null hypothesis of unit root in the series variables in first difference. Considering the case of additive outliers, in the face of structural break in the series variables in levels, the null hypothesis of a unit root in the series could not be rejected, in Table 10. The results in Table 11 indicate mixed results. Some series are unit root in levels but become stationary in first difference (BSL, INV and ASP).

Gregory-Hansen Structural Break Cointegration Test (Service, Agriculture and Manufacturing Sectors Value Added Model)-Model C; Model C/T; Model C/S and Model C/S/T)
Determinants of Service, Agriculture and Manufacturing sectors productivity growth are modeled using Gregory-Hansen (G-H) Co-integration approach in the presence of structural breaks The results on the estimated models revealed evidence of cointegration considering the various assumptions (a level shift; a level shift with a trend; a regime shift which allows the slope vector to shift as well and a regime shift which allows the slope vector to shift with a trend) using the ADF/Zt values. The models are estimated for the long run parameters for four explanatory variables due to the fact that Gregory-Hansen model takes a maximum of four variables.
The results on the estimated models [model C; model C/T; model C/S and model C/S/T] are reported in Table 11. The results revealed evidence of no significant cointegration in models C, C/T and C/S/T for service and manufacturing sectors but significant cointegration in model C/S. Models C/S/T and C/S are estimated for the service and manufacturing sectors. In the case or Agriculture sector, the results indicate that there is no significant cointegration in models C, and C/S and model C/S/T but significant cointegration in model C/T. Models C/S/T and C/T are therefore estimated for that sector.
The long run parameters of the two models selected for each of the three sectors are estimated using OLS. The results are presented in Table 19. The estimates of the models presents Model C/S as the only plausible model for service and manufacturing sectors and model C/T for Agriculture sector since some of the estimated coefficients of the explanatory variables are significant. The residuals from Model C/S and C/T are used to estimate the short run dynamic equation for the determinants of service, manufacturing and agriculture sectors, productivity growth with the error-correction adjustment model (ECM).   -8.780 1981 -8.880 1981 -57.030 1981 C -5.87 1993 -5.04 1992 -31.81 1992 C/T -5.73 1993 -5.06 1985 -31.52 1985 C/S -3.72 1999 -5.34 1987 -34.05 1987 C/T/S -6.41 1988 -5.42 1987 -34.54 1987

OLS: Using Observations 1970-2013 (T = 44) Dependent Variables: lnSSP, lnMSP and lnASP
The results of short-run dynamic equilibrium relationship coefficients estimated for the three sectors are reported in Table 13. All the explanatory variables in the models in respect of service and manufacturing sectors, with the exception of trade liberalisation are insignificant. The coefficients (5.97543) for SSP and (3.92482) for MSP of trade liberalisation have the expected priori theoretical sign. The error correction term (ECM) 0.877 for service sector is statistically significant; however it does not have the expected theoretical sign of negative. However, with MSP the error correction term (ECM) -0.475062 is statistically significant with the expected theoretical sign of negative. The results indicate that in the short-run financial liberalization influences the productivity growth of the manufacturing sector in Ghana.
In the case of agriculture sector, all the explanatory variables in the model are significant with the exception of Capital account openness. The error correction term (ECM) -0.957667 is statistically significant and has the ijbm.ccsenet.org International Journal of Business and Management Vol. 15, No. 4;2020 expected theoretical sign of negative. The implication is that in the short-run financial liberalization influences the productivity growth of the agriculture sector in Ghana.

Stability Tests
The results of the parameter stability tests are shown in Figures 3 4, 5, 6, 7, and 8. Figures 3 and 4   The results of the study in many respects confirm previous evidence (Didier & Schmukler, 2014, Larrain & Stumpner, 2013Dabla-Norris, Ho, & Kyobe;Schumpeter, 1911;McKinnon, 1973;Shaw, 1973;Goldsmith, 1969;Fry 1988;& Pagano, 1993). In the long-run, financial liberalization exercised a clearly positive impact on productivity growth of the economic sectors of Ghana, with a much stronger effect observed for service sector productivity growth than for manufacturing and agriculture sectors confirming the evidence found in Ghana by Jedwab and Osei (2012). The theoretical proposition that developing economies are closed economies with limited access to external finance and self-finance investments is confirmed Didier & Schmukler, 2014, Larrain & Stumpner, 2013Dabla-Norris, Ho, & Kyobe;Schumpeter, 1911;McKinnon, 1973;Shaw, 1973;Goldsmith, 1969); therefore alleviating financial restrictions in developing economies, is a means of exerting positive effect on growth in the sense that rising interest rates will lead to competitive market equilibrium, which will eventually promote efficient allocation of resources to productive sectors and industries for the purpose of enhancing productivity growth in Ghana.
The results obtained have empirical significance. There is a strong positive effect of trade openness on agriculture sector productivity growth and negative but insignificant effect on manufacturing and service sectors in the long-run. This result seems to reflect the relatively stronger involvement of agriculture activities in international trade relations that contributed to positive productivity-driven growth impulses. However, in the short-run there is a strong positive effect of trade openness on Service sector followed by manufacturing sector and significantly negative growth effect of Trade openness on agriculture sector productivity.
In the long-run Banking sector liberalization contributes to positive productivity-driven growth effects in service sector than manufacturing sector and significantly negative growth effect on agriculture sector productivity growth. In the short-run Banking sector liberalization exerts significantly negative growth effect on agriculture productivity growth and again negative but insignificant growth effect on manufacturing and service sector productivity growth. Capital account openness on the other hand contribute negatively to productivity growth in Observation CUSUMSQ plot with 95% confidence band ijbm.ccsenet.org International Journal of Business and Management Vol. 15, No. 4;2020 all the three sectors both in the long-run and short-run with the negative growth effects being significant for service sector and insignificant for manufacturing and agriculture sectors. There is also a strong positive effect of investment for manufacturing and service sectors productivity growth, however, the intensity of the respective influence differs, being stronger for manufacturing sector and weaker for service, and negative but insignificant effect on agriculture sector productivity growth in the long-run. Investment exerts positive growth effect on agriculture productivity growth in the short-run but insignificant growth effect on manufacturing and service sectors.
In the short-run however, the results of the study revealed mixed results because there is evidence from the study indicating both positive and negative growth effect of financial liberalization on the productivity growth of the three economic sectors in Ghana. A key reason that could explain the mixed results is that the effects of financial liberalization are time-varying (Schmukler & Kaminsky, 2007). The results of the study indicate that agriculture sector productivity growth is negatively influenced by trade liberalization in the short-run, on the contrary, trade liberalization exerts positive growth effects on both manufacturing and service sector productivity. However, in the long run, trade liberalization leads to a significantly positive impact on agriculture sector productivity growth as confirmed by Djokoto (2013). This result also support the evidence found by Schmukler and Kaminsky (2007) that financial liberalization is followed by large booms and busts during the short run period whiles in the long run, institutions improve and financial markets tend to stabilize. This means that reforms payoffs evolve over time and that productivity gains from reforms takes time to materialize (Dabla-Norris and others, 2015). The implication is that gains associated with financial reform programmes could be consolidated by avoiding drastic policy reversal, but rather concentrate on fine-tuning the existing policy measures and continued implementation of the financial sector reforms programmes

Conclusion
There is permanent effect of shock and structural breaks to financial liberalization and productivity growth of the service, manufacturing and agriculture sectors in Ghana. The theories of unit root in time series data are confirmed. The structural breaks coincide with identified climatic, economic, and political shocks in Ghana. The theoretical implications are that, the theories of structural breaks are supported, and that anticipated values of productivity growth that do not take into consideration the effects of structural breaks and external shocks might be affected by errors, and therefore exaggerated. Therefore, with regard to policy implication, government initiated structural reform that is not accompanied by sound macroeconomic policies may not have a lasting effect, since the effect of such reform in the long run growth path could be thrown out of gear by other external shocks.
The theoretical preposition that developing economies are closed economies with limited access to external finance and self-finance investments is confirmed; the implication is that alleviating financial restrictions in developing economies will lead to competitive market equilibrium, which will eventually promote efficient allocation of resources to productive sectors for promoting productivity growth in Ghana.
Theoretically, productivity growth model is confirmed. Wagner's hypothesis is also supported that government expenditure (investment in agriculture and manufacturing sector) affect productivity growth which in turn leads to economic growth. Expansionary fiscal policy is therefore supported as against contractionary policy. In Practice, increasing government spending (Investment), opening up trade, improving the financial sector will improve productivity growth of the economic sectors of Ghana. Considering the fact that financial liberalization exerts uneven effects on the productivity growth of the respective economic sectors of the economy, it is strongly recommended that policy makers should exercise caution in their choice of financial liberalisation indicators as a policy instrument for the purpose of realizing the intended productivity growth in the various sectors of the economy.
Financial liberalization exercises an uneven effect on the productivity growth of the various economic sectors of Ghana. The positive effect of financial liberalization on productivity growth can be broadly confirmed in all the three economic sectors (Agriculture, Manufacturing and Service Sectors). The results as revealed by the study are important in so far as they provide insights on the linkages between financial liberalization and productivity growth. To be precise, the results does not only confirm that financial liberalization might be growth-enhancing across the economic sectors, but they emphasize the evidence that such positive effects emanate from productivity improvements, which constitute an important channel through which developing countries could achieve economic growth.
Future studies could be focused on whether financial liberalization will make anticipated impact on sector level productivity growth using primary data from the various sectors of the economy in a survey study using structural modeling.
Time   1970197719841991