chapter onechss.uonbi.ac.ke/sites/default/files/chss/vol 4 no 1.docx · web viewalthough the...

DBA Africa Management ReviewApril 2014, Vol 4 No 1. Pp. 17-34

Outcomes of Worker Effort and Supervision in Tanzanian Labour Market

Godius Kahyarara, PhD.1

This paper examines the impact of work effort and supervision in the Tanzanian labour market. In particular it focuses on the extent to which observed earnings and productivity of a worker might be influenced by both the individual effort of a worker and intensity of supervision. To assess the earnings effect of work effort, the paper estimates the hourly earnings equation, which includes work effort and monitoring intensity among the determinants of the hourly earnings. The estimates control for unobserved firm specific effects and GMM production functions. Key findings of the paper are that a worker who exerts higher effort at work increases hourly earnings by about 27 per cent. Estimates of productivity affect via GMM shows that increase in the monitoring intensity increased the gross output per employee by about 34 per cent. The estimated coefficient is stable even after a range of factors are controlled for. The paper concludes that labour market reforms introduced in Tanzania on increased autonomy and flexibility of firm level work supervision and pay have positive outcomes for both employers and employees.

Key Words: worker effort. Labour market, Tanzania, Supervision, outcomesIntroductionThis paper examines the impact of work effort and work supervision in the labour market. In particular it focuses on the extent to which observed earnings and productivity of a worker might be influenced by both the individual effort of a worker and intensity of supervision. The paper focuses on Tanzania manufacturing. The labour market reforms introduced in Tanzania especially from the early 1990s increased autonomy and flexibility of firm level work supervision and reduced the role of state intervention in pay determination and work regulations. These changes may have had significant effects on the role of both work supervision and work effort in pay determination and productivity. Incentive pay theories for

example, predict that in setting wages employers face both adverse-selection and moral hazard problems: only workers know the difficulty of their jobs, and they can shirk so as to obscure this information from employers. Therefore, in the absence of state intervention in wage setting in Tanzanian labour market, firm level work supervision is one of the potential strategies for eliciting hidden actions and hidden information of the workers.

Although the changes introduced in the Tanzania labour market have potential effects on the level of earnings received by workers along with firm level productivity, little is known about the relationship between monitoring and worker earnings differences, or about the impact of worker effort on individual earnings and firm level

1 Lecturer, Economics Department, University of Dar-es-Salaam, Tanzania [email protected] |

DBA Africa Management Review


productivity. A better understanding of the structure and impact of internal supervision of work and worker effort is needed to understand their role in manufacturing development especially after the reforms. The goal of this paper is therefore to document several facts regarding work effort and intensity of work supervision in Tanzanian manufacturing.The overall objective of the paper is thus to examine the impact of worker effort on observed productivity and earnings. The specific objective of the paper is; To assess the extent to which indi-

vidual effort of the worker influence the level of productivity of labour in a particular firm

To analyse the link between observed earnings of an individual worker and the contribution he/she makes via the effort excreted at work

How a combination of the level of earnings and productivity is a function of the efforts of their employees.

The overall objective of the paper is thus to examine the impact of worker effort on observed productivity and earnings. The specific objective of the paper is; To assess the extent to which indi-

vidual effort of the worker influence the level of productivity of labour in a particular firm

To analyse the link between observed earnings of an individual worker and the contribution he/she makes via the effort excreted at work

How a combination of the level of earnings and productivity is a function of the efforts of their employees.

The paper acknowledges the fact that there are measurement problems asso-ciated with estimating the impact of

work effort and supervision heterogen-eity. For example, while higher effort can lead to higher earnings, it is also possible that higher earnings can be an incentive to elicit more effort. Like-wise, there is a possibility that the em-ployer can choose both the level of monitoring and earnings. If this is so then OLS estimates will be inconsist-ent. To address such problems, the pa-per estimates regression models that control for unobserved firm specific ef-fects and GMM production functions. The data used has information on methods of pay, i.e. time rate and piece rate. Such information provides poten-tial instruments that allow the paper to estimate two stage instrumental vari-able methods. After this introduction, the second section describes pay policy and work supervision in Tanzania over the period 1960-2012. In the third sec-tion theoretical framework underlying the assessment of earnings impact of pay method and work supervision is described. The discussion in this sec-tion reviews agency and efficiency wage theories that have been central to various models for analyzing pay method and work supervision. The Principal-Agent theory, shirking model of efficiency wage and other explana-tions of pay practices such as moral hazards, self -selection, and sociologi-cal explanations are briefly described in relation to work and supervision. The fourth section presents the models estimated in this paper along with pos-sible measurement problems. Methods employed to mitigate estimation prob-lems are also discussed in this section. Section five presents the estimation re-sults whereas the sixth section summa-rizes and concludes.

2 | DBA Africa Management Review


It is worth noting that, the 1990s have seen a reversal of many policies in Tanzania. Specific reform measures undertaken in the Tanzania labour market that may have direct impact on previous wage policy are: gradual elimination of fixed wages and introduction of flexible wage bargaining at enterprise level (both individually and collectively bargained wage); and allowing employers to provide fringe benefits (such benefits as food allowances, housing allowance, transport etc) to their employees when they find it appropriate (see Tsikata et al, 1999 Mjema et al, 1998, Mans, Darius 1994). Following the reforms in pay and work organization described above, several firms including the state owned ones, introduced performance based pay scheme, and most of the profitable firms increased wages substantially (Mans, Darius 1994)0. During pre-reform period, earnings differentials were low due to pay compression policy. The standard rate pay system, and absence of performance related pay and various constraints limited employers’ powers and authority over the employees. After the reforms, the current average earnings among the workers and across firms might differ significantly0. This study therefore addresses the question, whether and to what extent variation in earnings can be accounted for by heterogeneous pay method and supervision intensity during the post reform era. In the next section we describe the theoretical framework of our study.

Literature on Work and Efforts.

Previous studies in this area (see for example Strobl and Walsh, 2003) have argued that to correctly specify the impact of monitoring on earnings, the interaction between monitoring and work effort needs be modelled. The principal-agent framework (see for example, Ross 1973, Grossman and Hart 1983; and Sappington 1991 for details), is based on designing efficient explicit or implicit contracts between the principal (employer) and the agent (employee) in a situation where there are incentives for employees to cheat about their true work effort, largely arising from information asymmetries and monitoring costs. Related to the Principal-Agent explanation is the theory of implicit contracts (Azariadis (1975), Baily (1974) and Gordon (1974)). Based on this theory, if a firm is risk averse, it would like its workers to take a wage cut in bad states. However, if only the firm and not the workers can observe the state, wages cannot be made to depend on the state directly (Malcomsom, 1999). In the shirking model of Shapiro and Stiglitz (1984), workers can choose whether to work or shirk. Sociological explanations associated with George Akelof (1982) have also been a source of theoretical base for modeling monitoring, pay and earnings. According to Yellen (1984), sociological explanations can explain economic phenomenon related to pay such as reasons for forbidding piece rates even when feasible, or why workers might exceed firms set work standards.

In studies by (Lindberg and Snower (1987); Ewing and Payne (1999))

0 Performance based pay scheme was introduced in Tanzania Breweries Limited, Tanzania Cigarette Company, TPDC, Tanzania Electrical Company, IPP limited, Somaia group and others. 0 For example, Tanzania Breweries Limited raised the monthly minimum wage from Tanzania shillings 5,000 (about $ 16) in 1991 to 100,000 in mid 1990s and 186,000 recently (about $ 177) (Kileo, 2003). In addition to paying higher minimum wage, companies like Tanzania Breweries provide free medical treatment and free meals at lunch. 3 |



it is shown that workers earn more where monitoring is more difficult. Brown (1992); Ewing (1996); Parent (1999); Booth and Frank (1999) reveal that workers who are employed in jobs that have pay based on performance earn more. In other studies (such as Leonard (1987), Kruse (1992), Green and Mclntosh (1998), Groshen and Krueger (1990) etc), it is confirmed that there is a negative correlation between supervisory intensity and earnings. Such findings suggest that a worker will get more pay if he/she is less supervised. Yet, there is hardly any estimate of the earnings effect of monitoring intensity in Tanzanian labour market on a national level representative data for the recent period.

Theoretical Framework In this section the theoretical framework for estimating the effect of pay method and work supervision are described. Possible estimation problems that may be encountered during estimating the link between monitoring work supervision and earnings also form part of our discussion here. The theoretical frameworks underlying much of the previous work on the link between work effort, monitoring, earnings and productivity are based on agency theory (principal-agent framework), and efficiency wage theory, shirking model, adverse selection, moral hazard and sociological explanations.0 The principal-agent framework (see for example, Ross 1973, Grossman and Hart 1983; and Sappington 1991 for details), is based on designing efficient explicit or implicit contracts between the principal

(employer) and the agent (employee) in a situation where there are incentives for employees to cheat about their true work effort, largely arising from information asymmetries and monitoring costs. Employers are therefore forced to prepare contractual arrangements that are often self-enforcing, in the sense that they are designed to elicit the private information of the employees, providing them an incentive to provide their true efforts. The formal description of the Principal-Agent model is described below followed by other models from in the literature the area of pay method and monitoring.

Principal-Agent Model The models, consider a production function of an employee (the agent) such as; Q = X (e) + ε, [1]

Where e is the employee’s (agent’s) investment of effort, which is unobservable to the employer (principal), and X (e) defines expected output, which is strictly concave in effort, that is, X' (e) > 0 and X''(e) < 0, and where X (0) = 0. The second term, ε, is a random variable with zero mean. The variance of variable ε could be regarded as a measure of the risk of output. The employee’s (agent’s) utility, denoted as U, is a general function of his or her net payoff, P – C, [2]Where P is the realized wage and C is the cost of effort. The utility function is strictly concave in net payoff, that is, U '(P – C) > 0 and U''(P – C) < 0.

0 (see Ross 1973, Grossman and Hart 1983; Sappington 1991;Carlo Shapiro and Stiglitz, 1982; Steven Stoft, 1982; Samuel Bowles, 1981,1983;James Malcomson, 1981; Stiglitz, 1979b; Andrew Weiss, 1980;George Akerlof 1982; Edward Lazear (1981), Armen Alchian and Harold Dmsetz (1972), Stephen Ross (1973)).4 |



The cost function is strictly convex in effort, that is, C' (e) > 0 and C''(e) > 0, except at e = 0 where C´(0)= 0. The model assumes a linear pay contract for the agent, P = α+ βQ, where α is the fixed component of pay or insurance pay, and β the incentive parameter or pay-performance sensitivity. In the literature of managerial incentives, β is interpreted as incentive strength. The agent chooses effort to maximize his or her expected utility, E [U (P – C)], where E is the expectation operator, yielding the following incentive compatibility constraint:

βX'(e) – C'(e) = 0. [3]The constraint dictates that effort increases with the incentive parameter. The principal is risk-neutral in a competitive labor market and earns zero expected profit. Then, E (Q- P) = 0, or

α = (1 – β)X(e) [4]The optimisation problem of the model is to maximize the agent’s expected utility subject to the incentive compatibility constraint, equation (1), and the zero expected profit condition, equation (2). Before we describe how this framework will be useful for the analysis of this paper, we first describe other models that potentially link the pay method and monitoring with earnings.

Model Specification and DataThe purpose of this section is to outline the specification of the models to be estimated in analysing the effect of pay method and work supervision in our data. Also data available used in the study and the variables estimated are discussed. Two models are specified. First is the earnings function for assessing the effect of pay method, work effort and monitoring on earnings level. Second, the production functions for assessing the productivity effects of monitoring and effort are specified. The data used in this study is also described in this section. The discussion on data includes a detailed account of how various variables used in our estimation are created. The model specifications are given below.

Model SpecificationThe paper first specifies an earnings equation in which work effort and monitoring are among the regressors in the earnings function. The earnings equation [1] specified below include control variables for a range of firm and worker characteristics that may account for earnings differences.

Earnings FunctionLnW ijt=βo+ β 1 H ijt+β 2 Ε ijt+β 3 Μ jt+β 4 F ijt+β 5T ijt+π j+v ijt [1]

Where i, j and t are subscripts of individual, firm and time respectively, LnW is the log of real earnings of worker i in firm j during time t. Ei is the measure of work effort exerted by individual i. We

construct the measure of work effort from self-reported measurement of the level of effort supplied by the workers. We will describe the computation of this effort variable below. M is the measure of work monitoring intensity. This is calculated as



the percentage of managers and supervisor employed in a given firm. Hi is a vector of observable characteristics of worker i at time t (these characteristics could be human capital such as years of schooling, job tenure, work experience, job training and characteristics such as union membership etc), F is a vector of other control variables such as firm characteristics of ownership, location etc. The variable π represents unobservable characteristics i.e. omitted variables that may be correlated with explanatory variables in our earnings function and ν is the error term.

Production FunctionFor estimating the productivity effect of monitoring and work effort, a real a gross real output production function is specified in equation 2.

The gross real output production function is specified as follows;

LnQjt =α0 + α1LnKjt + α2LnLjt+ α3JTjt+ α4LnOHjt + α5LnRMjt + α6LnINDjt + α7Ejt +

α6Mjt + α8Cjt + µj + єjt

[2]

Whereby j and t are firm and time subscripts,

LnQ = log of real gross output

LnK = log of physical capital,

LnL = log of a number of labour available in a firm,

JT = variable for firm level job training,

OH = other human capital variables of weighted averages of schooling, age and tenure E = Variable for weighted Average Firm level EffortM = Variable for Firm Level Monitoring IntensityC = observable firm characteristics such as firm location, sector ownership, age export and others.LnRM = log of raw materials LnIND = log of indirect costs. µ = Firm Fixed effects and; є = error term.

Data and variables Data source is the Tanzanian manufacturing firm surveys over a long span of time. The data were originally collected under the Regional Program of Enterprise Development (RPED) surveys by World Bank, later continued by the University of Oxford and later the University of Dar-es-Salaam. Work effort variable is measured using self-reported measure of the effort level exerted at a worker level. This is derived from a psychological question asked during the data collection. In particular, a worker is asked “How tired are you at the end of the day? Each worker is expected to choose one out of the following four answers: 1) very tired, (2) tired (3) not really tired (4) not tired at all. Using these four responses we construct a zero-one dummy variable of effort that takes the value of one if a worker is either very tired or tired and zero otherwise. Specifically, if a worker feels very tired or tired then is regarded as “High Effort worker” and Low Effort worker if he feels not really tired or not tired at all. We denote this variable as E in the model [1] above.



The weighted average of work effort within a firm is derived from firm level information about individual level of self-reported effort. To arrive at the firm level effort variable, the individual level effort value is weighted by the proportion of workers in a given occupational category in a firm. Earnings variable is the hourly rate of earnings obtained using information on the number of hours worked and the total current earnings received. The hourly earnings include salary/wage, plus any allowances received. Other variables are as defined in the previous papers.

Empirical ResultsThis section presents empirical results based on the earnings and production functions. The paper first estimate the earnings effect of work effort and monitoring from earnings function, and then present the results of the estimates of productivity effect of these variables using the production function. The OLS results are presented first and then consideration of other options of estimates mentioned earlier, of the firm effect within regressions and GMM (for the productivity effects) is effected. While the earnings effects of work effort and monitoring are presented in section 5.1, we present the productivity effect of these two variables in section 5.2.

Estimates of earnings effect of pay methods and monitoring on earnings In estimating the results, the key question here is whether there is any evidence of earnings effect of high work effort in our data. This hypothesis is supported by the results. As the results in column two of table 1, controlling for human capital, there are earnings differences accounted

for by the fact that a worker exerts higher effort. The estimated coefficient on work effort is 0.269. It is highly significant at 1 percent. This coefficient has positive sign suggesting that a worker who exerts higher effort at work increases hourly earnings by about 27 per cent. Therefore, the results support the hypothesis that controlling for the pay method and human capital those who work hard are paid more. Hence, the monitoring level will influence the earnings effect of work effort. We therefore move to the third column and see how monitoring in our data influences hourly earnings. The major hypothesis tested using the results in column [3] are whether high monitoring can reduce the necessity of paying higher earnings. The results reported in this third column strongly accept this hypothesis. There is evidence of a negative correlation between monitoring intensity and hourly earnings in our data. The coefficient on monitoring is 0.554. It has a negative sign suggesting that highly monitored workers are less paid.

To account for the effect of such influence in our estimation we interact the work effort with monitoring level. The results of earnings function that estimate the earnings effect of the interaction between work effort and monitoring are presented in column 4 of table 1. The results indicate a negative effect of the interaction between work effort and monitoring on the individual effect of monitoring on earnings, although the coefficient estimate is not statistically significant. Specifically, we find that after we control for the joint effect of the interaction between work effort and monitoring, the coefficient size of monitoring does not change, while the statistical significance is reduced by half



although is still significant. We interpret this as some evidence that the earnings effect of monitoring work through work effort.

However, work effort and the level of monitoring are likely to be influenced by worker characteristics and firm characteristics. Our estimates in columns [1-4] do not control for other factors that might be picked up by the pay methods and monitoring coefficients. For this reason, the paper stepwise adds the control variables of firm and worker characteristics in columns [5-6] respectively. The Tanzania Manufacturing Enterprise Survey from which our data is drawn contains various firm and individual characteristics such as location, sector, occupation, and ownership. The estimates first add job classification, sector

ownership and location in column [5]. Moving to the right hand side of the table, it is certainly clear that the earnings effect of work effort and monitoring operate through these other factors. For instance it is confirmed that controlling for occupation and other firm characteristics in column [5] significantly changes the coefficient size of all our variables of interest. The coefficient estimate of work effort is more than halved, and becomes weakly significant. Such findings suggest that the earnings effect of work effort work through other characteristics such as occupation. In column [5] the estimates control for observable characteristics that have potential influence on or earnings effect of work effort and monitoring. Evidences of earnings effect of monitoring are still observed.

Table 1: Firm level hourly earnings effect of work effort and monitoring in Tanzanian manufacturing

OLS1 OLS2 OLS3 OLS4 OLS5 FFE-MYears of Education -0.034 -0.026 -0.032 -0.031 -0.028 -0.019

(1.87) (1.44) (1.74) (1.73) (1.56) (1.20)Education Squared 0.006 0.005 0.006 0.006 0.005 0.004

(4.74)** (3.82)** (4.19)** (4.14)** (3.58)** (3.09)**Currently Job training 0.016 0.019 0.001 0.001 0.018 -0.099

(0.28) (0.34) (0.02) (0.02) (0.33) (1.80)Past Job Training 0.123 0.124 0.117 0.117 0.085 0.012

(3.11)** (3.14)** (2.99)** (2.99)** (2.19)* (0.34)Age 0.074 0.072 0.069 0.069 0.064 0.056

(6.81)** (6.64)** (6.47)** (6.47)** (6.05)** (5.69)**Age squared -0.001 -0.001 -0.001 -0.001 -0.001 -0.001

(6.19)** (6.08)** (5.97)** (5.97)** (5.62)** (5.04)**Years of Tenure 0.001 0.001 0.000 0.000 0.001 0.004

(0.50) (0.23) (0.15) (0.14) (0.39) (1.52)High Work Effort 0.251 0.267 0.267 0.108 0.091

(5.61)** (6.01)** (6.01)** (1.52) (1.52)Monitoring Intensity -0.564 -0.530 -0.917 -0.692

(5.73)** (2.81)** (4.47)** (2.07)*Effort*Monitoring -0.044 0.330 0.303

(0.21) (1.46) (1.47)Administration -0.081 -0.200

(0.65) (1.88)Cleric -0.345 -0.384

(2.88)** (3.81)**Sales -0.436 -0.481

(3.18)** (4.10)**Supervisor -0.227 -0.418

(1.90) (4.06)**



Technical -0.374 -0.427(2.96)** (3.90)**

Production -0.510 -0.591(4.20)** (5.69)**

Apprentices -0.731 -0.615(4.34)** (3.95)**

________________________________________________________________________________________________________________CONTROL VARIABLESLocation NO NO YES YES YES YESOwnership NO NO NO YES YES YESSector NO NO NO NO YES YESFirm Fixed Effects NO NO NO NO NO YESObservations 1821 1821 1821 1821 1821 1821R-squared 0.32 0.34 0.35 0.35 0.37 0.26

Absolute values of t-statistics are in parentheses. Significance at the 1 per cent, 5 per cent and 10 per sent level is indicated by ***, ** and * respectively.



Estimates of the Impact of Work Efforts and Monitoring on Firm Level ProductivityThis section uses firm level measures of work monitoring and effort to test if there is any evidence of productivity effect of these variables. It does so by estimating the firm level gross output production functions in which weighted average work effort and firm level monitoring intensity of a firm are used as the determinants of firm level productivity. The results of the estimates of productivity outcomes of monitoring and efforts are presented in table 2 below. To begin with the estimate checks if constant returns to scale are accepted. The test results for constant returns to scale reported in column [1] of Table 2 indicates that the constant returns to scale is rejected. Further findings are that weighted average work effort has positive effect on firm level productivity measured by gross output. The estimated coefficient on average work effort is 0.21. It has a positive sign implying that a 1 per cent increase in average work effort of the firm will increase gross output by 0. 21 percent. However, the results are weakly significant (t=1.34). The coefficient on monitoring has a negative sign but its level of statistical significant is very low (t=0.89). Therefore, based on these results, controlling for production inputs and human capital, there are weak evidences of positive effect of work effort on productivity.

The second column, add exports and firm age variables. The results show that when export and firm age are controlled for, there is slightly decline in the coefficient estimate and the statistical significance. But there is still evidence that average work effort of a firm has a

positive effect on firm level productivity measured by gross output. In the third and fourth columns, more firm characteristics and time invariant unobserved characteristics are controlled for respectively. The results show that, when a broad range of firm characteristics is controlled for both the coefficient size and statistical significance of the firm level effort variable falls substantially. In particular, it is found that control for the firm fixed effects eliminates the statistical significance the effort variable (as t falls to 0.36). Further results are that the productivity impact of monitoring increases substantially. The estimated coefficient is now positively correlated with productivity and the statistical significance nearly doubles although still insignificant (t=1.39). The results suggest that the productivity effect of work effort work and monitoring partly work through unobserved firm fixed effects. Thus the OLS productivity effect of work effort observed in the OLS gross output production function might be picking up these unobserved firm fixed effects.

The results reported in tables 2 are based on OLS hence do not control for endogeneity problem we discussed above. To account for this problem, the paper adds estimate the gross output production functions using GMM. The results are reported in table 3. Based on the results, the estimates based on the gross output per employee in table shows a significant positive impact of monitoring on gross output per employee. The coefficient estimate is 0.34. It suggests that increase in the monitoring intensity increases the gross output per employee by about 34 per cent. The estimated coefficient is stable even after a range of factors is controlled



for. The results partly reflect the size effect. The level of monitoring intensity decreases with firm size. Therefore an in

gross output per employee is likely to be a increased efficiency of monitoring.

Table 2: OLS estimate results of the impact of work effort on firm level gross real output

OLS1 OLS2 OLS3 FE (Within)Log of Capital 0.016 0.018 0.015 0.009

(1.28) (1.36) (1.12) (0.07)Log of labour 0.167 0.170 0.185 0.127

(6.61)** (6.72)** (7.47)** (1.44)Log of Raw Materials 0.649 0.649 0.646 0.703

(25.98)**(26.28)**(26.44)**(17.46)**Log of indirect Cost 0.198 0.192 0.185 0.092

(6.81)** (6.59)** (6.55)** (1.76)Weighted Average Effort 0.210 0.207 0.187 0.084

(2.34) (2.31) (2.19) (1.46)Weighted Average Monitoring 0.061 0.052 0.044 0.024

(1.89)* (1.75)* (1.63)* (1.39)Weighted Past Training 0.044 0.067 0.062 -0.056

(0.59) (0.91) (0.84) (0.47)Weighted Current Training -0.123 -0.127 -0.103 0.581

(1.10) (1.12) (0.93) (2.45)*Average Years of Education -0.004 -0.005 -0.002 0.025

(0.45) (0.53) (0.19) (1.84)Average Years of Tenure 0.006 0.006 0.007 -0.012

(1.67) (1.70) (1.79) (1.97)Average Years of Experience 0.005 0.005 0.005 0.004

(1.77) (1.78) (1.67) (0.79)Exports 0.049 0.043 0.042

(0.93) (0.81) (0.38)Firm Age -0.001 -0.001 -0.005

(1.18) (1.02) (0.11)______________________________________________________________________________________CONTROL VARIABLESLocation NO NO YES YESOwnership NO NO YES YESSector NO NO YES YESFirm Fixed Effect NO NO NO YESObservations 297 297 297 297R-squared 0.99 0.99 0.99 0.84CRS1test Σßi=1(p-value) 0.034 0.050 0.040 0.64

Absolute values of t-statistics are in parentheses. Significance at the 1 per cent, 5 per cent and 10 per sent level is indicated by ***, ** and * respectively. CRS test is an F-test for constant returns to scale that the coefficients on inputs sums to unity. The weighted average of schooling, tenure, job training and age are derived from firm level information about individual highest level of education completed, the occupational specialization, work tenure whether an individual attended job training and experience. Each value is weighted by the proportion of workers in a given occupational category in each firm to obtain a weighted average for each firm. The occupational categories included are managers, administration, sales, clerical supervisor, technicians, production workers and support staff.



Table 3: GMM estimate results of the impact of work effort on firm level gross real output per employee

GMM1 GMM2 GMM3 GMM4Log Capital Per Employee 0.132 0.158 0.143 0.118

(1.58) (1.70) (1.53) (1.89)Log of Raw Materials Per Employee 0.699 0.662 0.678 0.669

(7.91)** (6.37)** (8.15)** (9.56)**Log of Indirect Cost 0.066 0.072 0.078 0.103

(0.90) (0.95) (0.92) (1.64)Weighted Average Past Training 0.021 0.063 0.075 0.115

(0.13) (0.41) (0.49) (0.71)Weighted Average Current Training -0.130 -0.148 -0.160 -0.251

(0.57) (0.68) (0.72) (0.99)Average Years of Education 0.002 0.001 0.004 -0.003

(0.19) (0.07) (0.33) (0.18)Average Years of Tenure 0.003 0.003 0.003 0.003

(0.55) (0.45) (0.64) (0.50)Average Years of Experience 0.009 0.009 0.008 0.007

(2.12)* (1.95) (2.16)* (2.00)*Weighted Average Work Effort 0.144 0.121 0.101 -0.056

(0.67) (0.52) (0.43) (0.22)Firm Level Monitoring Intensity -0.344 -0.391 -0.359 -0.370

(1.79) (1.90) (1.57) (2.25)*Exports 0.009 -0.034 -0.114

(0.07) (0.35) (1.31)Firm Age -0.005 -0.004 -0.002

(1.56) (1.48) (0.79)Observations 172 172 172 129Robust z-statistics in parentheses

______________________________________________________________________________________CONTROL VARIABLESLocation NO NO YES YESOwnership NO NO YES YESSector NO NO YES YESFirm Fixed Effect NO NO NO NO2000 0.000 0.000 0.000 0.000Observations 114 114 114 114CRS1test Σßi=1(p-value) 0.27 0.25 0.42 0.38

Absolute values of t-statistics are in parentheses. Significance at the 1 per cent, 5 per cent and 10 per sent level is indicated by ***, ** and * respectively. CRS test is an F-test for constant returns to scale that the coefficients on inputs sums to unity. The weighted average of schooling, tenure, job training, work effort and age are derived from firm level information about individual highest level of education completed, the occupational specialization, work tenure whether an individual attended job training and experience. Each value is weighted by the proportion of workers in a given occupational category in each firm to obtain a weighted average for each firm. The occupational categories included are managers, administration, sales, clerical supervisor, technicians, production workers and support staff.



ConclusionThis paper set out to assess the impact of work effort and work supervision on earnings and productivity in Tanzanian manufacturing over the period 1990-2012. The paper makes use of self- reported measures of effort to test if there are any gains from working hard. The OLS based finding is that high effort workers are compensated for exerting more effort at work. In particular, the results show that a worker who exerts higher effort at work increases hourly earnings by about 27 per cent. The paper argues that such results support the hypothesis that controlling for the pay method and human capital those who work hard are paid more. In estimating the earnings effect of monitoring, the results are evident that controlling for human capital monitoring intensity reduces earnings by 0.554. The argument partly reflect size effect but most interestingly are consistent with the incentive theories which suggest that loosely monitored workers receive more real earnings than highly monitored ones. Connected to this finding the paper confirms earnings effect of monitoring work through work effort. However, the results based on firm fixed effects regressions have revealed that to a large extent the OLS based effects are influenced by the unobserved time invariant firm attributes. We argued that this reflected that the effects picked up by OLS are plagued with these effects. Estimates of productivity affect via GMM shows a significant positive impact of monitoring on gross output per employee. The coefficient estimate suggests that increase in the monitoring intensity increased the gross output per employee by about 34 per cent. The estimated coefficient is stable even after a range of

factors are controlled for. The level of monitoring intensity decreases with firm size. Therefore an increase in gross output per employee is likely to be a reflection of efficiency in production from effective monitoring. The paper concludes that labour market reforms introduced in Tanzania on increased autonomy and flexibility of firm level work supervision and pay have positive outcomes for both employers and employees. This is so because prior to the centrally controlled pay system significantly reduced the ability of manufacturing firms to design establishment level motivation such as performance based and productivity enhancing schemes. Hence any increase in work effort or monitoring cold not affects pay or incentives to work more.

ReferencesAkerlof, G. and J. Yellen (1988),"Job

Switching and Job Satisfaction in the U.S. Labour Market," Brookings Papers on Economic Activity

Akerlof, George, (1982) " Labour Contracts as Partial Exchange," Quarterly Journal of Economics, 543-69.

Altonji, Joseph and Shakotko, Robert, (1987) “Do Wages Rise with Job Seniority?” Review of Economic Studies, 54, July 1987, 437-59.

Baker, G. (1992). “Incentive Contracts and Performance Measurement.” Journal of Political Economy, 100(3):598-614.

Becker, Gary,(1962) Human Capital, (Chicago: University of Chicago Press).

Bishop, John, (1987)“The Recognition and Reward of Employee Performance,”



Journal of Labour Economics, 5 (Part 2), S36-S56.

Borjas, George J., (1981) “Job Mobility and Earnings Over the Life Cycle,” Industrial and Labour Relations Review, 34, 365-76.

Bulow, J., and Summers, Lawrence, (1986) “A Theory of Dual Labour Markets with Application to Industrial Policy, Discrimination and Keynesian Unemployment,” Journal of Labour Economics, 4, 376-415.

Brown, C. (1990). “Firms’ Choice of Method of Pay.” Industrial and Labour Relations Review, 43(3):165s-182s.

Booth, Alison L., and Frank, Jeff. (1996) “Seniority, Earnings, and Unions.” Economica 63 : 673-86.

Brown, Charles. (1990) “Firms’ Choice of Method of Pay.” Industrial and Labour Relations Review 43, special issue: S165-S182.

Ehrenberg, R. (1990). “Do Compensation Policies Matter?” Industrial and Labour Relations Review, 43(3):

Gibbons, R. (1987). “Piece-Rate Incentive Schemes. Journal of Labour Economics, 5(4):413-29.

Goldin, C. (1986). “Monitoring Costs and Occupational Segregation by Sex: A Historical Analysis.” Journal of Labour Economics 4(1):1-27.

Goldin, Claudia. (1986) “Monitoring Costs and Segregation by Sex: A Historical Analysis.” Journal of Labour Economics 4 : 1-27.

Hutchens, Robert M., (1987) “A Test of Lazear’s Theory of Delayed Payments Contracts” Journal of Labour Economics, 5 (Part 2) S153-S169.

ILO/JASPA, (1982). “Basic Needs in Danger” ILO Addis Ababa.

ILO (1990) Employment, Unemployment, and Underemployment. An ILO manual on concepts and methods. Geneva.

ILO/JASPA (1989). African Employment Report 1988. ILO. Addis Ababa.

Kanemoto, Y., and W. Macleod. (1992). “The Ratchet Effect and the Market for Second-hand Workers.” Journal of Labour Economics, 10(1):85-98.

Kusilati, T.M. (1996) Formal Sector Employment Reduction and Trade Union Responses, ILO Paper Presented at a National Seminar held in Dar es Salaam, April 1996.

Knight, J and Sabot, R. (1990) Education, Productivity and Inequality. The East African Natural Experiment. World Bank Publication, Oxford University Press.

Kruse, Douglas, (1992) “ Supervision, Working Conditions, and the Employer Size-Wage Effect,” The Journal of Industrial Relations, 31, 229-249

Lazear, E., and S. Rosen. (1981). “Rank-Order Tournaments as Optimum Labour Contracts.” Journal of Political Economy 89(5):841-64.

Lazear, Edward P. (1979) “Why is There Mandatory Retirement?” Journal of Political Economy, 87, December 1979, 12611-84.



Lazear, Edward P. ( 1981) “Agency, Earnings Profiles, Productivity and Hours Restrictions,” American Economic Review, 71, 606-20.

Lazear, Edward P and Moore, Robert L. (1984) “Incentives, Productivity and Labour Contracts,” Quarterly Journal of Economics, 99, 275-96.

Lazear, Edward. “Salaries and Piece Rates. (1986) “Journal of Business 59: 405-32.

Leonard, Jonathan. (1987). “Carrots and Sticks: Pay, Supervision, and Turnover.” Journal of Labour Economics, Vol. 5, pp. s136-152.

Lester, Richard. (1952). “A Range Theory of Wage Differentials.” Industrial and labour Relations Review, Vol. 5, pp. 483-499.

Macleod, W., and J. Malcomson. (1989). “Implicit Contracts, Incentive Compatibility, and Involuntary Unemployment.” Econometrica 57(2):447-80.

Malcomson, James M., (1984) “Work Incentives, Hierarchy, and Internal Labour Markets,” Journal of Political Economy, 92: 486-507.

Milgrom, P., and J. Roberts. (1992). Economics, Organization, and Management. Engle-wood Cliffs, N.J.: Prentice Hall.

Mitchell, Daniel J.B., Lewin, David and Lawler, Edward, (1990 “Alternative Pay Systems, Firm Performance, and Productivity,” from Paying for Productivity: A Look at the Evidence, ed. Alan S. Blinder (Washington, D.C.: The Brookings Institution): 15-94.

Mazumder,D (1993), "Labour Markets and Adjustment in Open Asian Economies: The Republic of Korea and Malaysia." World Bank Economic Review, Vol.7 No.3

-------------(1994),"The Structure of Wages in African Manufacturing", RPED Paper no. 33. The World Bank. September.

Mans, D. (1994),"Tanzania Resolute Action”, in I. Hussain and R. Faruqee (eds). Adjustment in Africa, Lesson from country case studies. WB, 1994, PP 352-426.

Mjema and Shitundu, (1996) "Employment and labour markets during adjustment: The case of Tanzania": a consultancy report submitted to ILO J.L.M. / ILO; ERB. -- Dar es Salaam.

Mazumdar,D. (1994), wages in Africa (draft), (Washington DC, World Bank).

Morin .P., Artus, P., & F Legendre,(1991), Le partage implicite des propfits et ses effets sur la productivité , Annales d’Economie et de Statistiques, 22, pp 33-57.

Mwongozo (1971): The Party Guidelines Printpak, Dar es Salaam, United Republic of Tanzania.

Osterman, Paul, (1994) “Supervision, Discretion, and Work Organization,”



American Economic Review Papers and Proceedings, 84, 380-84.

Pencavel, John.(1997) “Work Effort, On-the-Job Screening, and Alternative Methods of Remuneration.” In Research in Labour Economics, vol. 1, edited by Ronald Ehrenberg, pp. 225-58. Greenwich, CT: JAI. Performance-Related Pay.

Polachek, Solomon W. and Siebert, W. Stanley, (1993) The Economics of Earnings, (Cambridge: Cambridge University Press) .

Seiler, E. (1984). “Piece Rate vs. Time Rate: The Effect of Incentives on Earnings.” Review of Economics and Statistics 66(3):363-76.

Shapiro, Carl, and Stiglitz, Joseph, (1984) “Equilibrium Unemployment as a Worker Discipline Device,” American Economic Review, 74, 433-44.

Shearer, B. (1996). “Piece Rates, Principal-Agent Models, and Productivity Profiles.” Journal of Human Resources 31(2):275-303.

Stevens, M. (1994), “ public Expenditure and Civil Service Reform in Tanzania” in Rehabilitating Government: Pay and Employment Reform in Africa. World Bank, Washington D.C.

Stiglitz, J. (1975). “Incentives, Risk, and Information: Notes Towards a Theory of Hierarchy.” Bell Journal of Economics 6(2):552-79.

Strobl. E and Walsh. F (2003), “Estimating the Shirking Model with Variable Effort. Discussion Paper series No 75. CORE, Universite Catholique de Louvin, Belgium.

Tsang M C (1987) The Impact of underutilization of education on productivity: A case of US Bell companies. Economics of Education Review 6 (3): 239-54

Tsikata and Wangwe (1999) Macroeconomic developments and Employment in Tanzania, Summary of ILO/MLYD ESRF paper.

Weitzman, Martin. (1985) “The Simple Macroeconomics of Profit-Sharing.” American Economic Review 75: 937-53.



The Mediation Effect of Customer Perception on the Relationship Between Quality Drivers and Customer Satisfaction in Large Maize Flour Mills in

Nairobi, Kenya

Ngungu kabare, PhD.1, Francis N Kibera, PhD.2 and Justus M Munyoki, PhD.3

This study sought to establish the influence of quality drivers on the satisfaction of direct business customers within large Maize Flour Mills in Nairobi and assess the mediation effect of customer perception on this relationship. The quality drivers studied were product quality, service quality, complaints handling, ease of doing business and product price. Customer perception constructs studied were customer’s desire for features critical to quality, brand imagery, firm imagery and reference to competitive substitutes. Primary data were collected in February 2013 by use of questionnaires from 81 direct Business Customer firms randomly selected from 13 Maize Flour Mills in the study area grinding at least 15 MT of maize per day. Results showed that the influence of quality drivers on customer satisfaction is both direct and partially mediated by customer perception, both influences being positive and statistically significant (p< 0.01). Quality of service significantly influenced customer satisfaction (β= 0.441, p< 0.01) and most of the other quality drivers and intention to recommend. Brand imagery significantly influenced satisfaction (β= 0.531, p< 0.01) followed by desire for features critical to quality (β= 0.259, p< 0.01). These results have implications for marketing theory. The finding that customer perception partially mediated the process of customer satisfaction agrees with the consumer attitude theories which postulate that attitude and subjective norms in conjunction with cognitive and emotional considerations influence intentions which in turn give impetus for action. The study contributes to the evolution and adaptation of customer satisfaction models by adding customer perception as mediator variables. Further, the results have implications useful at national policy level. Kenya’s strategy for revitalizing agriculture and vision 2030 both aspire to increase the country’s regional and global trade through improved efficiency and competitiveness at firm level, agro-processing and the marketing system including the wholesale and retail sectors. The volume of trade within the East African Community is expected to increase as member states reduce trade barriers. This will open new trade opportunities but could increase competition. Training local firms on the issues of quality drivers and customer perception can help to improve their regional and global competitiveness. For managerial practice, the results demonstrate that frequent feedback on customer perception is necessary and that improvements in the quality of service go a long way in improving customer perception concerning other quality drivers and satisfaction. It is concluded that customer satisfaction enhancement programs and evaluation models need to integrate primary drivers of quality with key drivers of customer perception. The study was limited in a number of ways. Due to time, cost and other constraints a cross-sectional research design was used and focused on firms in Nairobi. Data were collected from respondents once to get their views and perceptions concerning a limited number of variables and constructs. However, perceptions vary over time and across markets or regions as influenced by changes in consumer preferences or economic changes that influence purchase and consumption patterns. Opportunities therefore exist for longitudinal and wider studies in the same area of research.

1 Lecturer in Marketing, School of Business, Kenyatta University2 Professor in Marketing, School of Business, University of Nairobi3 Senior Lecturer in Marketing, School of Business, University of Nairobi17 |



Key Words:Quality drivers, customer perception, customer satisfaction (CS), Nairobi, Kenya

IntroductionCustomer satisfaction refers to the overall evaluation of how pleasurable one’s interaction with an organization is including the buying and use experience, relative to what is anticipated (Anderson et al, 1997; Kotler and Keller, 2006; Ronald, 2010). At higher levels satisfaction can lead to loyalty which is a deeply held commitment to repurchase or re-patronize a preferred product consistently over time even with stiff competition (Oliver, 1999). Satisfaction is a moving target that is often shaped by various attributes upon which customers form perception. The attributes could be related to the product such as quality, value-price relationship, benefits and features, design, reliability and consistency and product/ service range. They can be related to service such as delivery, complaint handling and problem resolution. Other attributes are related to the buying process such as convenience, courtesy, communication, staff competence and firm reputation (Crawford, 2007; Dutka, 1993).

Kenya’s manufacturing sector contributes 10% of the country’s GDP. Food processing contributes about two thirds of the manufacturing GDP and about a fifth of the country’s export earnings (Osano et al, 2008). Trade in maize flour plays a key role in this sector, especially because maize meal is the staple food in the country (Wangia et al, 2002). On average households in Nairobi spend 27% of their food budget on staples with maize meal taking the lead (Kamau et al, 2011).

However, other carbohydrate sources such as wheat, rise, potatoes, sorghum and

cassava are gaining popularity (Kamau et al, 2011; Muyanga et al, 2005). New challenges have come by way of legislation such as the VAT Act 2013 that has moved cereal milling byproducts and several other commodities from zero rating to the standard VAT rating of 16%. This will increase the cost of these supplies and is likely to affect demand thereby increasing competition (Deloitte, 2011). In view of this, research on the dynamics of quality drivers of the maize flour would be useful in policy issues related to the country’s strategy of promoting agro-processing and food security. In a study on Kenyan urban consumption of maize meal, Mukumbu and Jayne (1994) found that the key quality drivers on purchase decisions were price and convenience followed by taste and nutritive value. The current study sought to establish the effects of customer perception on the relationship between quality drivers and customer satisfaction within large maize flour mills in Nairobi.

Literature ReviewPerception relates to how individuals see the world around them. It is a process by which people select, organise, and interpret stimuli into meaningful and unified pictures or images of situations (Schiffman and Kanuk, 2007). Through a positive image a consumer perceives a brand or a firm to be stable, dependable and suitable for satisfying the needs of the consumer. Such an image can strengthen a firm’s credibility, lead to more sales and help to fight competition. Consequently many companies strive to develop, project and maintain positive images of their brands and the firm. Both the brand and corporate image reinforce one another in



that if the brand image is positive, it reflects favourably on the corporate image and vice versa (Haedrich, 1993).

Satisfaction shifts as customer perception of quality changes, evolves, and grows to encompass more expectations (French et al, 2005). Issues can arise concerning established brands, product features, processes or procedures including complaints. Competitors can offer better alternatives or changes in other fields such as technology or culture can shift customer perception (Armitage and Conner, 2001; Ferrell and Hartline, 2005). Customers develop expectations depending on how they perceive attributes and base decisions on perceptions rather than on the basis of objective reality (Schiffman and Kanuk 2007). The relative importance played by respective quality drivers and other enabler variables in fostering customer satisfaction varies over time as marketing conditions and other aspects of life change. This dynamism needs to be reflected in satisfaction assessment tools and associated frameworks if they are to remain robust in capturing the voice of the customer. The scope and nature of drivers used in satisfaction models therefore needs to be reviewed from time to time so as to keep abreast of changes in consumer behaviour and related fields (Johnson et al, 2001).

Satisfaction is closely linked to future purchase behaviour and willingness to recommend and is thus a strong predictor of loyalty and customer retention (Ferrell and Hartline, 2005; Turkyilmaz and Ozkan, 2007). Satisfaction therefore helps to reduce customer turnover and lower transaction costs related to contract negotiations, order processing, and

bargaining (Fornell, 1992). Satisfied customers are most likely to share their experiences with about five or six people while a dissatisfied customer is more likely share their unfortunate experience with up to ten people (Ronald, 2010). Loyal customers tend to buy more, are less price sensitive, speak well of the firm and are harder for competitors to win (Schiffman and Kanuk, 2007).

Many firms in Kenya are increasingly focussing on enhancement of customer satisfaction due to increasing competition. An increasing number are registering with industry and global quality standards such as ISO 9001:2008 Quality Management System and, for food related operations, the FAO/WHO Codex Alimentarius HACCP Food Safety System. Assessment of CS is a key feature of these standards (Hashim, 2007; Kimbrell, 2000). This way firms hope to compete more effectively locally and against imports as well as in the export markets. Anyango and Wanjau (2011) observed improved company performance in Nairobi with respect to perceived quality, competitive advantage, corporate image and market share associated with adoption of ISO 9001 certification. Furthermore the certification impacted positively on financial resource management (p=0.001) and customer satisfaction (p=0.03).

MethodologyThis study used a descriptive cross-sectional design to check for significant associations between the study variables and make generalisations concerning the target population. The population of study comprised all direct Business Customers in the sifted maize flour sector within the administrative boundaries of Nairobi City.



These included distributors, wholesalers, supermarkets, and other institutions that bought maize flour directly from the maize mills. With the help of the Maize Flour Mills’ marketing and sales managers, a random sample of 10 direct Business Customers was selected from the customer data bases of each of thirteen maize mills in the study area grinding at least 15 MT of maize per day. Primary data were collected by use of a semi- structured questionnaire. Respondents were purchasing managers because they interact with the mills and are responsible for the flour sourcing function.

A total of 81 questionnaires were received back out of the 130 questionnaires sent out. Rating was done on a ten point scale (ranging from 1 to 10) to increase the level of scale details. Pearse (2011) and Preston and Colman (2000) report that rating scales with less than seven points tend to have inadequate granularity. They obtained the most reliable scores from scales with seven to ten response categories. Likewise Reichheld (2003)

observes that scales with more points offer wider options especially because cstomers tend to refrain from top scores.Data were cleaned, edited and coded followed by analysis and reporting. The statistical programme Software Package for the Social Sciences (SPSS) version 12.0 was used to analyse the data using both descriptive and inferential statistics. Normality of distribution was checked through skewness and kurtosis tests. Correlations were used to examine variable relationships. Simple and multiple linear regressions were used to test for the study hypotheses. The coefficient of determination (R2) indicated the amount of variation explained by the model. Mediation was tested in accordance with the four steps regression procedure described by Baron and Kenny (1986) and Fairchild and MacKinnon (2009). Figure 1 shows the mediation path diagrams next to the conceptual model in which X is the independent variable (quality drivers), M the mediating variable (customer perception) and Y the dependent variable (customer satisfaction).


XM

Y

Conceptual model

X

M

Y

Mediation paths

ba

cl

H3

X = Independent variable; Y= Dependent variable; M= Mediating variable; a, b, and c = beta coefficients

H1

H4

H2


Figure 1: Mediation model path diagram

Reference for mediation paths: Fairchild and MacKinnon (2009), A General Model for Testing Mediation and Moderation Effects. Prevention Science, 10:87-99.

In step one of mediation testing, the dependent variable Y was regressed on the independent variable X and the standardized regression coefficient (beta for path c) examined to determine the size and direction of the relationship and checked for significance. This beta for path c was significantly different from zero and therefore in step two, the mediator M was regressed on the independent variable X to estimate the standardized beta regression coefficient for path a, which was examined to determine the size and direction of the relationship and was significantly different from zero. In step three, Y was regressed on M to determine the beta coefficient for path b, which was significant. In step four, the

dependent variable Y was regressed on X while controlling the effect of M on Y, by performing a hierarchical regression analysis that placed M and X in successive independent variable boxes in the SPSS program. If both coefficients for paths a, and b are significant, then M mediates the relationship between X and Y and cl is assessed to check the link strength (Fairchild and MacKinnon, 2009; Bennett, 2000; Shaver, 2005; Sharma et al, 1981).

FindingsCorrelationsService quality emerged as the feature with the most profound positive influence on other quality drivers and on satisfaction and intention to recommend. Service



quality had influence on product quality (r = 0.46, p< 0.05) and moderate correlations with complaints handling and ease of doing business as well. It had a positive non-significant correlation with product price. The implication of this is that

Business customers are likely to associate good service with superior quality of product and associated processes. Figure 2 highlights the correlations between the level of service performance and the other attributes.

Figure 2: Correlation Coefficients- service quality and other parameters

Satisfaction

Recommend

ProductQuality

BusinesEase

ComplaintHandling

Price0

0.050.1

0.150.2

0.250.3

0.350.4

0.450.5

0.294

0.455 0.460.427

0.269

0.148

Cor

rela

tion

with

ser

vice

leve

l

`

** p< 0.01, * p< 0.05, ns: not significant, N= 81, Source: Primary Data

The highly statistically significant correlation between service performance and intention to recommend (r= 0.46, p< 0.01) implies that the higher the level of service performance experienced by a business customer, the higher the likelihood that the customer will recommend the brand or firm to a colleague or friend. The highly statistically significant correlation coefficient relating to ease of doing business (r= 0.427, p< 0.01) suggests that a high level of service performance experienced by a business

customer reassures the client that any problem arising from the purchase will be attended to expeditiously.

Correlation coefficients between customer perception and quality drivers (Figure 3) were positive and statistically significant for all quality drivers except for price whose coefficient fell slightly outside the threshold of p< 0.05. The correlation coefficients were also positive and statistically significant for both overall satisfaction and intention to recommend.


** ** ** ** * ns


Figure 3: Correlations Between Quality Drivers and Customer Perception

00.05

0.10.15

0.20.25

0.30.35

0.40.45

0.50.374000000000

003

0.236 0.238

0.384000000000003

0.281 0.2660.208

Cor

rela

tion

with

Cus

tom

er P

erce

ptio

n

`

** p< 0.01, * p< 0.05. Source: Primary Data

The correlation coefficients were highly statistically significant (p< 0.01) for service quality and overall satisfaction. The managerial implication is that any improvements on attribute performance (quality drivers) translates to improved positive customer perception about the Flour Brand or Mill and that good quality service has a particularly profound effect on enhancing customer perception.

The results further indicated that customer perceptions had a positive and statistically significant influence on customer satisfaction (r= 0.374, p< 0.05) and

intention to recommend (r= 0.236, p< 0.05). This implies that sellers stand to gain more on customer satisfaction by focussing on both product attributes and customer perception as opposed to working on the attributes alone.

Mediation (Intervening) Effects of Customer PerceptionThe study had hypothesized that customer perception mediates the relationship between quality drivers and customer satisfaction as shown in Figure 4.


** * ***** ns

Customer Satisfaction

QualityDrivers

CustomerPerception

IV

MV

DV

Step 2 Step 3

Step 1

Step 4 (H4) is done if the first 3 steps are statistically significant and predictors are entered at successive blocks in SPSS program to check for change in R2. In this fourth step DV is regressed on IV while controlling for the effect of MV

IV= independent variable, MV= mediating variable, DV= dependent variable

(H1)

(H2)(H3)


Figure 4: Mediation Testing Steps

In step 1 (hypothesis H1) customer satisfaction was regressed on quality drivers and the relationship was positive and statistically significant (β = 0.391, B = 0.406, p< 0.01) and the model accounted for 15.3% of the variation. This supported the first condition for testing the effect of

mediation. This hypothesis tested the direct relationship between quality drivers and customer satisfaction and was stated as shown below. Aggregate mean scores of CS were regressed against those of the quality drivers. The output is shown in Table 1 a to c.

H1: There is a statistically significant relationship between quality drivers and customer satisfaction.

Table 1: Customer Satisfaction regressed on aggregate mean scores of Quality Drivers

a) Model SummaryModel R R2 Adjusted R2 Std. Error of the Estimate1 .391(a) .153 .142 1.08776

a Predictors: (Constant), Quality Drivers

b) ANOVA(b)Model Sum of Squares Df

Mean Square F Sig.

1 Regression 16.827 1 16.827 14.222 .000(a) Residual 93.475 79 1.183 Total 110.302 80

a Predictors: (Constant), Quality Drivers, b Dependent Variable: C. Satisfaction



c) Coefficients(a)

Model Unstandardized Coefficients

Standardized Coefficients T Sig.

B Std. Error Beta 1 Constant 4.831 .836 5.778 .000 QualityDrivers .406 .108 .391 3.771 .000

As shown in Table 1 there was a statistically significant linear relationship between quality drivers and customer satisfaction (β = 0.391, B = 0.406, p< 0.05) and hence the study failed to reject hypothesis H1. The influence of quality drivers on customer satisfaction was moderate as the model accounted for 15.3% variability (R2 = 0.153). The resulting simple linear regression model that can be used to predict the level of satisfaction for a one standard deviation improvement in the performance level of quality drivers can be expressed as:

CS = 4.831 + 0.391QD + e ............................................................... (1)

Where CS = level of customer satisfaction and QD = level of quality drivers performance. The standardized beta coefficient 0.391 implies that, other factors constant, a one standard deviation

improvement in the performance of quality drivers would raise the level of customer satisfaction by a factor of about 0.4 of a standard deviation.

Step 2 (hypothesis H2) involved assessing whether quality drivers predicted customer perception and whether the relationship was statistically significant. The hypothesis was stated as follows:

H2: There is a statistically significant relationship between quality drivers and customer Perception.

In this second step aggregate mean scores of customer perception were regressed on those of quality drivers and the relationship was positive and statistically significant (β = 0.418, B = 0.590, p<0.01) and the model explained 17.4% of the variation, supporting the second condition for mediation testing as presented in Table 2 a to c. The study failed to reject hypothesis H2.

Table 2: Customer Perception regressed on Quality Drivers

a) Model Summary

ModelR R Square

Adjusted R Square

Std. Error of the Estimate



1 .418(a) .174 .164 1.45732a Predictors: (Constant), Quality Drivers

b) ANOVA(b)

Model Sum of Squares Df

Mean Square F Sig.

1 Regression 35.426 1 35.426 16.681 .000(a) Residual 167.779 79 2.124 Total 203.205 80

a Predictors: (Constant), Quality Drivers, b Dependent Variable: Customer Perception

c) Coefficients(a)

Model Unstandardized Coefficients


B Std. Error Beta 1 Constant 2.146 1.120 1.916 .059 QualityDrivers .590 .144 .418 4.084 .000

a Dependent Variable: Customer Perception. Source: Primary Data

The resultant regression model that predicts the level of customer perception for a given level of quality drivers’ performance can be expressed as:

CP = 2.146 + 0.418QD + e, ................................................................(2)

where CP = customer perception and QD = quality drivers.

The model shows that for one standard deviation improvement in the performance of quality drivers, customer perception would improve by 0.418 of a standard deviation.

Step 3 (hypothesis H3) involved checking whether customer perception predicted

customer satisfaction and whether the relationship was statistically significant. The hypothesis was stated as follows:

H3: There is a statistically significant relationship between customer perception and customer satisfaction (CS)

In this third step of mediation testing, aggregate mean scores of customer satisfaction were regressed against those of customer perception and the relationship was positive and statistically significant (β = 0.349, B = 0.257, p<0.01) and the model explained 12.2% of the variation, supporting the third condition for mediation testing as shown in Table 3 a to c. The study failed to reject hypothesis H3.



Table 3: Customer Perception Predicting CS

a) Model SummaryModel R R2 Adjusted R2 Std. Error of the Estimate1 .349(a) .122 .111 1.10713

a Predictors: (Constant), Customer Perception

b) ANOVA(b)Model Sum of Squares Df Mean Square F Sig.1 Regression 13.469 1 13.469 10.989 .001(a) Residual 96.833 79 1.226 Total 110.302 80

a Predictors: (Constant), Customer Perception, b Dependent Variable: Customer Satisfaction

c) Coefficients(a)Model

Unstandardized Coefficients


B Std. Error Beta 1 Constant 6.233 .533 11.701 .000 Customer Perception .257 .078 .349 3.315 .001

a Dependent Variable: Customer Satisfaction. Source: Primary Data

The resulting regression model that predicts the level of customer satisfaction (CS) for a given level of customer perception (CP) is:

CS = 6.233 + 0.349CP + e ............................................................................................(3)

The model indicates that for a unit standard deviation improvement in the level of customer perception about a brand or firm, customer satisfaction level would improve by a factor of about 0.349 of a standard deviation.

The success of the first three conditions for mediation testing lead to the conduct of the final test in line with hypothesis H4 which was stated as follows:

H4: Customer perception has a mediating effect on the relationship



between quality drivers and customer satisfaction.

Customer satisfaction was regressed on quality drivers while controlling for the effect of customer perception to check for the significance of the resultant R2 change and coefficients for quality drivers. Statistical insignificance would imply full

mediation otherwise it would be partial (Baron and Kenny, 1986; Shaver, 2005). Customer perception was loaded into block two in SPSS program to control for its effect. Both the R2 change (R2= 0.073) and the coefficient (β= 0.296) were statistically significant (p<0.05) indicating partial mediation. Results are shown in Table 4.

Table 4: Satisfaction regressed on Quality Drivers while controlling for Customer Perception

Model Summary

Model R R2 Adj R2

Std. Error of Estimate Change Statistics

R2

ChangeF Change

df1

df2

Sig. F Change

1 .349-a .122 .111 1.10713 .122 10.989 1 79 .0012 .441-b .195 .174 1.06720 .073 7.022 1 78 .010

a Predictors: (Constant), Customer Perception, b Predictors: (Constant), Customer Perception, Quality Drivers

a)ANOVA(c)

Model Sum of Squares Df

Mean Square F Sig.

1 Regression 13.469 1 13.469 10.989 .001(a) Residual 96.833 79 1.226 Total 110.302 80 2 Regression 21.467 2 10.734 9.424 .000(b) Residual 88.835 78 1.139 Total 110.302 80

a Predictors: (Constant), Customer Perception, b Predictors: (Constant), Customer Perception, Quality Drivers, c Dependent Variable: Satisfaction

b)Coefficients(a)

Mod

el Unstandardized Coefficients


B Std. Error Beta


Customer Satisfaction

H1: β = 0.391**, R2 =0.153 direct effect

H3, β =0.349**H2, β =0.418**

QualityDrivers

CustomerPerception

IV

Mediating Variable

DV

Step 2

Step 1

Step 3R2 =0.174 R2 =0.122

(Step 4= H4: β = 0.296* for QD, β =0.226* for CP, R2 change=0.073* mediated effect)where QD = quality drivers, CP = customer perception


1 Constant 6.233 .533 11.701 .000 Customer Perception .257 .078 .349 3.315 .0012 Constant 4.474 .839 5.332 .000 Customer Perception .166 .082 .226 2.018 .047 Quality Drivers .308 .116 .296 2.650 .010

a Dependent Variable: Satisfaction. Source: Primary Data

The study failed to reject H4. The resulting regression model from the fourth step of mediation testing, expressed on the beta coefficients can be expressed as:

CS = 6.233 + 0.296QD + 0.226CP + e ............................................................(4)

Figure 5 shows a diagrammatic summary of the results for mediation testing.

Figure 5: Summary Results of Mediation Effect Testing

**P< 0.01, * p < 0.05, β = beta coefficient, IV = Independent Variable, DV = Dependent Variable. Source: Primary Data

The four regression equations relating to the tests for mediation effect, expressed in beta coefficients are:

Step 1: CS = 4.831 + 0.391QD + e Step 2: CP = 2.146 + 0.418QD + e

Step 3: CS = 6.233 + 0.349CP + eStep 4: CS = 6.233 + 0.296QD + 0.226CP + e

where CS= customer satisfaction, QD= quality drivers, CP= customer perception



The bigger beta coefficient relating to quality drivers (β= 0.296) compared to that of customer perception (β= 0.226) in step 4 implies that, other factors constant, business clients probably place slightly more emphasis on quality drivers. However, sellers need to foster improvements in both quality drivers and customer perception as both influence customer satisfaction.

DiscussionThis study sought to establish the influence of quality drivers on the satisfaction of business customers within large Maize Flour Mills in Nairobi and assess the mediation effect of customer perception on this relationship. Quality of service emerged as a key precondition for customer satisfaction. This agrees with the indications of satisfaction index models and the Service Profit Chain that service quality is a key motivator in customer satisfaction (Heskett et al, 1994; Anderson et al, 1997; Fornell et al, 1996). It also agrees with the findings of Silvestro and Cross (2000) who observed a strong positive correlation between service quality and customer satisfaction at the 95% level and concluded that a key aim of management should be to improve perceptions of service quality for their customers. Ramaseshan and Vinden (2009) reported that quality drivers accounted for up to 54% of satisfaction with retail stores. Simmerman (1995) reported that 70% of customer desertions were due to poor service compare to 20% combined for price and product quality. Adams (2006) found that employee attitude was often a leading cause of customer defections (68%) followed by other dissatisfactions (14%) and desertions (9%).

The results showed that the influence of quality drivers on customer satisfaction is positive and statistically significant (p< 0.05) and is partially mediated by customer perception. Flour Mills therefore need to routinely survey on customer attitudes. Brand image had positive and statistically significant effect on customer satisfaction (β= 0.513, p< 0.05). This agrees with the attitude theories which postulate that attitude and subjective norms in conjunction with cognitive and emotional considerations influence intentions which in turn give impetus for action (Bagozzi, 1992; Batra et al, 1996). Macinnis and Price (1987) reported that imagery influences cognitive, physiological, and behavioural responses. They further observed that imagery has a positive influence on incidental learning and given that much of consumer learning is incidental, then it is likely that imagery influences likelihood and timing of purchasing.

ConclusionThe objectives of this study were to evaluate the influence of quality drivers on the satisfaction of business customers in the sifted maize flour sub sector in Nairobi and to assess the mediation effect of customer perception on this relationship. Both the direct and the mediated relationships were positive and statistically significant (p< 0.05). The findings therefore supported the two main hypotheses of the study. In addition, quality of service had stronger positive correlations with customer satisfaction and intention to recommend compared to other quality drivers. Customer imagery of the brand and the firm emerged as a key driver of customer perception.



It is concluded that the influence of quality drivers on customer satisfaction is both direct and partially mediated through customer perception. This implies that Flour Mills need to actively pay attention to the direct quality drivers such as product quality and price as well as customers perception variables such as user imagery of the brand and firm. Improvements in the quality of service go a long way in enhancing customer satisfaction. The study calls for the incorporation of dimensions of customer perception in customer satisfaction evaluation index models and satisfaction surveys.

ImplicationsThe findings from this study have implications for the theory of marketing, policy and managerial practice. The finding that customer perception partially mediated the process of customer satisfaction agrees with the consumer attitude theories which postulate that attitude and subjective norms in conjunction with cognitive and emotional considerations influence intentions which in turn give impetus for action. The study contributes to the evolution and adaptation of customer satisfaction models by adding customer perception as a mediator variable. Most models of customer satisfaction focus mainly on primary quality drivers (Johnson et al, 2001).

The results have policy implications that could be harnessed to promote competitiveness. The volume of trade within the East African Community is expected to increase as member states reduce trade barriers (KPMG, 2013). This will open new trade opportunities but could increase competition. Kenya’s strategy for revitalizing agriculture and

vision 2030 both aspire to increase the country’s regional and global trade through improved efficiency and competitiveness at firm level, including agro-processing and across the marketing system (Ministry of Agriculture, 2004; Ministry of Planning, 2007). Training, research and development are key ingredients of the strategy and this can include the need to consider quality drivers and perception in satisfaction variables.

For managerial practice, the results demonstrate that business customers within the Maize Flour subsector are more willing to do business with Flour Mills that offer superior service quality, are efficient in resolving complaints and have positive brand imagery. Improvements on the quality of service influence customer satisfaction both directly and through positive influences on other drivers of quality. They further suggest that satisfaction surveys need to collect feedback on both the quality drivers and customer perception including ratings relative to competitors.

Areas for Further ResearchAs this was a cross-sectional research that studied customer satisfaction dynamics in a sector at a particular point in time, other research could use longitudinal research design to track changes over time. Rust et al., (1999) reported that besides mere quality limits, perceived variability and/ or consistency in quality over time is important to capture as well. Such deeper insights on dynamics of quality drivers would help marketers and brand managers in a competitive market such as the local maize flour subsector to refine their market offerings and customer satisfaction



programmes for a better competitive advantage.

Secondly more variables can be studied and wider geographical coverage. Extra variables can include the increasing availability of alternative carbohydrate sources in Kenya, the growing use of fortified flour blends and increasing dietary consciousness ((Mukumbu and Jayne, 1994; Muyanga et al., 2005). These trends are likely to lead to changes in the consumption of maize flour which would limit generalization of study findings for forecasting and estimation. Furthermore one can disaggregate quality attributes along the Kano model’s ‘critical to quality’ dimensions with a view to identifying the key performance factors that often form the common basis for competition (Anderson and Mittal, 2000; Oliver 1999).

ReferencesAdams C.W. (2006). Customer satisfaction

and customer loyalty are the best predictors of customer retention. www.adamssixsigma.com.

Anderson E.W., and Mittal V. (2000). Strengthening the Satisfaction- Profit Chain. Journal of Service Research, 3 (2), 107-120

Anderson E.W., Fornell C., and Rust R.T. (1997). Customer satisfaction and profitability: Differences between goods and services. Marketing Science, 16 (2), 129-145.

Anyango D. M., O., and Wanjau K. (2011). Assessment of the Relationship between ISO 9001 Certification and Performance of Manufacturing Firms in Nairobi. Paper presented at the 2nd

African International Business and

Management (AIBUMA) conference at KICC, Nairobi, 24th to 26th August, 2011.

Armitage J. C., and Conner M. (2001). Efficacy of the Theory of Planned Behaviour: A meta-analytic review. British Journal of Social Psychology. 40, 471-499.

Bagozzi R.P., (1992). Self-Regulation of Attitudes, Intentions, and Behaviour. Social Psychology Quarterly, 55 (2), 178-204.

Baron R. M., and Kenny D. A. (1986). The Moderator- Mediator Variable Distinction in Social Psychological Research: Conceptual, and Statistical Considerations. Journal of Personality and Social Psychology, 51 (6), 1173-1182.

Batra R., Myers J.G., and Aaker D.A. (1996). Advertising Management 5th Ed. Prentice Hall, India, page 210.

Bennett J.A. (2000). Mediator and Moderator Variables in Nursing Research: Conceptual and Statistical Differences. Research in Nursing & Health, 23, 415-420.

Crawford R. (2007). Customer Priorities: what customers really want. www.instituteofcustomerservice.com

Deloitte (2011). The Value Added Tax Bill 2011. Page 13. Deloitte and Touche. www.deloitte.com

Dutka A. (1993). The AMA Handbook for Customer Satisfaction. www.esurveyPro.com

Fairchild A. J., and MacKinnon D. P. (2009). A General Model for Testing


http://www.esurveyPro.com/

http://www.deloitte.com/

http://www.instituteofcustomerservice.com/

http://www.adamssixsigma.com/


Mediation and Moderation Effects. Prevention Science, 10, 87-99.

Ferrell O.C., and Hartline M.D. (2005). Marketing Strategy, 3rd Ed. Thompson South-Western, pages 10, 94-95.

Fornell C. (1992). A National Customer Satisfaction Barometer: The Swedish Experience. The Journal of Marketing, 56, (1), 6-21.

Fornell C., Johnson M.D., Anderson E.W., Cha J., and Bryant B.E. (1996). The American Customer Satisfaction Index: Nature, Purpose and Findings. Journal of Marketing, 60, (4), 7-18.

French P. D., Sutton S., Hennings J. S., Mitchell J., Wareham J. N., Griffin S., Hardeman W., and Kinmonth L. A. (2005). The Importance of affective beliefs and attitudes in the theory of Planned Behaviour: Predicting intention to increase physical activity. Journal of Applied Social Psychology. 35, (9), 1824-1848.

Haedrich G. (1993). Images and strategic corporate and marketing planning. Journal of Public Relations Research. 5(2), 83-89.

Hashim M. (2007). Quality Management System ISO 9001:2000. www.ukm.my

Heskett J.L., Jones T.O., Loveman G.W., Sasser E.W., and Schlesinger L.A. (1994). Putting the Service-Profit Chain to Work. Harvard Business Review March-April, 164-174.

Johnson M.D., Gustafsson A., Andreassen T.W., Lervik L., and Cha J. (2001). The evolution and future of national customer satisfaction index models. Journal of Economic Psychology, 22, 217-245.

Kamau M., Olwande J., and Githuku J., (2011). Consumption and Expenditure on Key Food Commodities in Urban Households: The Case of Nairobi. Tegemeo Institute of Agricultural Policy and Development. Working paper No. 41.

Kimbrell E. (2000). What is Codex Alimentarius?. AgBioForum – Volume 3, Number 4 – 2000 – Pages 197-202. www.agbioforum.org.

Kotler P., and Keller K.L. (2006). Marketing Management. 12th Ed. Pages 25-26, 194.

KPMG (2013). Kenya 2013 Budget Brief. Page 6. KPMG Kenya. www.kpmg.com

Macinnis D.J., and Price L.L. (1987). The Role of Imagery in Information Processing: Review and Extensions. Journal of Consumer Research, 13, 473-491.

Ministry of Agriculture (2004). Strategy for Revitalizing Agriculture 2004 – 2014 Government Printers, Kenya.

Ministry of Planning (2007). Kenya Vision 2030. Pages 13-14. Government Printers, Kenya

Mukumbu M., and Jayne T.S. (1994). Urban Maize Consumption Pattern: Strategies for Improving Food Access for Vulnerable Urban Households in Kenya, in Proceedings of the Symposium on Agricultural Policies and Food Security in East Africa, East Africa Association of Agricultural Economists, May 19-20, 1994.

Muyanga M., Jayne T.S., Kodhek A., and Ariga J., (2005). Staple Food Consumption Patterns in Urban Kenya: Trends and Policy Implications. Tegemeo Institute of Agricultural Policy


http://www.kpmg.com/

http://www.agbioforum.org/

http://www.ukm.my/


and Development. Working paper No. 19.

Oliver R.L. (1999). Whence Consumer Loyalty? Journal of Marketing, 63, 33-44.

Osano E., Kiuluku P., Njuguna H., and Tutuka L. (2008). Round table presentation on the manufacturing industry in Kenya. www.roundtableafrica.net.

Pearse N. (2011). Deciding on the Scale Granularity of Response Categories of Likert type scales: The Case of 21- Point Scale. The Electronic Journal of Business Research Methods, (9), (2), 159-171.

Preston C., and Colman A. M. (2000). Optimal number of response categories in rating scales: reliability, validity, discriminating power, and respondent preference. Acta Psychologica, 104, 1-15.

Ramaseshan B. and Vinden A. (2009). Drivers of Retail Store Loyalty. Australian and New Zealand Marketing Academy, www.duplication.net.au

Reichheld, R.F. (2003). The One Number You Need to Grow. Harvard Business Review, December, pages 46-56.

Ronald G. (2010). The American Customer Satisfaction Index (ACSI), The European Performance Satisfaction Index, The NCSB, www.van-haaften.nl

Rust R.T., Inman J.J., Jia J., and Zahorik A. (1999). What you don’t know about customer perceived quality: The role of

customer expectation distributions. Marketing Science, 18 (1), 77-92.

Schiffman L.G., and Kanuk L.L. (2007). Consumer Behaviour 9th Ed. Prentice Hall India. Pages 9, 35, 241-247.

Sharma S., Durand R.M., and Gur-Arie O. (1981). Identification and Analysis of Moderator Variables. Journal of Marketing Research, 18, 291-300.

Shaver J.M. (2005). Testing for Mediating Variables in Management Research: Concerns, Implications, and Alternative Strategies. Journal of Management, 31 (3), 330-353.

Silvestro R. And Cross S. (2000). Applying the service profit chain in a retail environment- challenging the ‘satisfaction mirror.’ International Journal of Service Industry Management, 11, (3), 244-268.

Simmerman S. (1995). Using Dis-Un-Empowerment to Build Customer and Employee Loyalty. www.squarewheel.com

Turkyilmaz A., and Ozkan C. (2007). Development of a customer satisfaction index model. An application to the Turkish mobile phone sector. Industrial Management and Data Systems. 107 (5), 672-687.

Wangia C., Wangia S., and Groote H. (2002). Review of maize marketing in Kenya: Implementation and impact of liberalization, 1989-1999. Proceedings of the 7th E & S. Africa Regional Maize Conference, Nairobi, Kenya 11-15 Feb 2002.


http://www.squarewheel.com/

http://www.van-haaften.nl/

http://www.duplication.net.au/

http://www.roundtableafrica.net/


Turn of The Month Effect: Evidence From The Nairobi Securities Exchange

Ogilo Fredrick1

This study sought to investigate if Turn of the Month effect exists at the Nairobi Securities Exchange. In carrying out the study, the days of the month were divided into two, the Turn of the Month (TOM) which included the last trading day of the month and the first three trading days of the following month. The other trading days of the month were categorized as Rest of the Month (ROM). The 20 share index was used as the sampling frame and the daily indices were used to compute the daily returns. Secondary data was obtained from the Nairobi Securities Exchange data base. The TOM coefficient was not significant to confirm TOM effect. It is therefore concluded that there is no TOM effect at the Nairobi Securities Exchange. To practice, the study will give vital information to brokerage firms as they will advise their clients on the best time of the month to sell or buy securities. The findings of the study will also be of benefit to policy formulation aimed at improving capital market efficiency.

Keywords: Turn of the Month, Rest of the Month, Twenty Share Index, Market Efficiency and Nairobi Securities Exchange.

IntroductionIt was Fama (1970) who defined an efficient market as a market which adjusts rapidly to new information. He categorized the capital market in to three different forms of efficiencies namely, the weak form, the semi-strong form and the strong form efficiencies. Rubinstein (1975) extended the definition of market efficiency. He said that the market is

efficient with regard to an informational event if the information causes no portfolio changes. Kennedy (1996) added that, markets in the developing and less developed countries are not efficient in semi-strong or strong form but in the weak form.

There has been criticism to the efficient Market Hypothesis that has generated

1 Lecturer, School of Business, Mombasa Campus, University of Nairobi, Kenya [email protected]



concern. Grossman and Stiglitz (1980) argue that perfectly information efficient markets are impossible, for if markets are perfectly efficient, the return to gathering information is nil, in which case there would be little reason to trade and markets would eventually collapse. Campbell, Lo and MacKinlay (1997) share the same opinion. They are in favour of the notion of relative efficiency, that is, the efficiency of one market measured against another.

Lo and MacKinlay (1999) argue that Efficient Market Hypothesis, by itself, is not a well-defined and empirically refutable hypothesis. To make it operational, one must specify additional structure, including investors’ preferences, information structure and business conditions. But then a test of the Efficient Markets Hypothesis becomes a test of several auxiliary hypotheses as well, and a rejection of such a joint hypothesis tells us little about which aspect of the joint hypothesis is inconsistent with the data. The Bad Model problem advanced by Fama (1991) suggests that Efficiency per se is not testable. That it must be tested jointly with some model of equilibrium. Financial anomalies have been advanced in the EMH theory. A financial anomaly refers to unexplained results that deviate from those expected under the financial theory. Financial anomalies that have been identified under the EMH include, Low Price Earnings Effect, low priced stocks, the small firm effect, the neglected firm effect and Market overreaction. Others are the Turn of the calendar anomalies which include The January effect, the day of the week effect, the turn of the month effect, the turn of the year effect among others. The study narrows down to turn of the

month effect and will seek to establish if it exists at the Nairobi Securities Exchange.

Turn of the Month EffectA turn-of-the-month effect is documented by Ariel (1987) where higher mean stock returns occur during the initial days of a trading month than during days later in the month. Ariel (1987) was the first to identify this anomaly in US stock prices at the beginning of one month and end of the other month. He studied this effect by considering last day of one month and the first three days of upcoming month. Changes in stock prices in these days are found positive.

Zafar, Shah and Urooj (2009) observe that different studies have given different conclusions for on TOM effect. Cadsby (1989) carried a study which confirmed that the turn-of-the-month effect is present in both US and Canada. Further researches were carried out by Cadsby and Ratner (1992) and the results suggest that the anomaly has shown its presence in the US, Canada, the UK, West Germany, Switzerland and Australia but not in Japan, France, Italy and Hong Kong.

MethodsThe study employed an analytical research design. The design was preferred due to the fact that the study entailed analyzing returns during the TOM and ROM windows to establish the window with the highest return and so confirm if TOM effect exists or not. For purposes of this study the population consisted of all active firms listed at the NSE for equity trading as at December 2011. The study used the NSE 20 share index. Secondary data from the NSE database was used for years ranging from 2002 to 2011. The regression



model below was used to determine TOM effect:Rt = β0+β1d2t+εt

Where:Rt = Daily return of stock index

β0 = is the coefficient of ROMβ1 = is the difference in the computed mean returns of TOMd2t = Dummy variable for the TOM returnsεt = Error term

Results And DiscussionsTable 1: Regression coefficients for TOM effect for individual years from 2002 to 2011 (t-values are in parenthesis)Periods β0 β1 R P-value

2002 -0.035(-0.606)

0.193(1.469)

0.009 0.143

2003 0.252(3.124)

0.397(0.790)

0.003 0.430

2004 0.77(1.466)

-0.175(-2.095)

0.17 0.037

2005 0.116(3.007)

0.136(0.239)

0.000 0.811

2006 0.144(0.101)

0.138(-0.002)

0.000 0.999

2007 -0.035(3.124)

0.068(0.631)

0.002 0.529

2008 -0.327(-3.002)

0.462(3.178)

0.039 0.002

2009 -0.026(-0.376)

0.058(-0.2020

0.000 0.840

2010 0.123(2.713)

0.123(0.013)

0.000 0.989

2011 0.220(40.20)

0.243(1.868)

0.10 0.063

2002-2011 0.510(0.359)

0.150(0.304)

0.000 0.761

Source: Research dataThe model regression results in table 1 above have shown the coefficient TOM β1 to be insignificant except for two years 2004 and 2008. In year 2008, TOM coefficient is negatively significant with β1 = -0.175, P-value = 0.037 which is less

than 0.05 and t-value = -2.095. In year 2008 TOM is positively significant with β1 = 0.462 where P-value is less than 0.5, that is 0.002 < 0.5 with t-value of 3.178. TOM coefficient for the whole period of ten years is also insignificant thus failing to demonstrate TOM effect at the NSE.



Table 2: Year 2002 Model SummaryModel

RR Square

Adjusted R Square


Change StatisticsR Square Change

F Change df1 df2

Sig. F Change

dimension0 1 .093a .009 .005 .81029 .009 2.159 1 246 .143a. Predictors: (Constant), Days of the month

Source: Research data

The coefficient of determination (R2) equals 0.009. This shows that days of the month explain 0.9 percent of the total variation of Daily return of stock index leaving only 99.1 percent unexplained.

The P- value of 0.143 implies that the model of Daily return of stock index is not significant at the 5 percent significance level.

Table 3: ANOVAModel

Sum of Squares df Mean Square F Sig.1 Regression 1.418 1 1.418 2.159 .143a

Residual 161.516 246 .657Total 162.934 247

a. Predictors: (Constant), Days of the monthb. Dependent Variable: Daily return of stock index

Source: Research dataThe P- value of 0.143 implies that there is no linear regression relationship between days of the month and the Daily return of stock index at 5 percent significance level. That is, the Daily return of stock index in the year 2002 during ROM (zero value of x) was -0.035 and the Daily return of stock index was expected to increase at the rate

of 0.193 during TOM (value of x equal 1). The P- value of 0.143 > 0.05 implies that, Days of the month as an independent variable is not significant at the 5 percent. TOM coefficient (0.228) is higher than ROM coefficient (-0.035) but not significant to confirm TOM effect in year 2002.


RR Square

Adjusted R Square



F Change df1 df2

Sig. F Change

dimension0 1 .050a .003 -.002 1.14430 .003 .624 1 247 .430a. Predictors: (Constant), Days of the month




In table 4 above the coefficient of determination (R2) equals 0.003. This shows that days of the month explain 0.3 percent of the total variation of Daily return of stock index leaving only 99.7

percent unexplained. The P- value of 0.43 implies that the model of Daily return of stock index is not significant at the 5 percent significance level.

Table 5: ANOVAModel

Sum of Squares Df Mean Square F P-value1 Regression .817 1 .817 .624 .430a

Residual 323.425 247 1.309Total 324.242 248

a. Predictors: (Constant), Days of the month


The P- value of 0.43 in fig. 4.3.2C above implies that there is no linear regression relationship between days of the month and the Daily return of stock index in year 2003 at 5 percent significance level. That is, the Daily return of stock index in the year 2003 during ROM (zero value of x) was 0.252 and the Daily return of stock index was expected to increase at the rate

of 0.145 during TOM (value of x equal 1). The P- value of 0.43 > 0.05 implies that, Days of the month as an independent variable is not significant at 5 percent. Though the TOM coefficient is more than ROM coefficient, it is not significant to confirm existence of TOM effect at NSE for year 2003.

Table 6: Year 2004 Model Summary

Model

RR Square

Adjusted R Square



F Change df1 df2

Sig. F Change


Source: Research dataThe coefficient of determination (R2) in table 11 equals 0.017. This shows that days of the month explain 1.7 percent of the total variation of Daily return of stock index leaving only 98.3 percent

unexplained. The P- value of 0.037 implies that the model of Daily return of stock index is not significant at the 5 percent significance level.

Table 7: ANOVAModel

Sum of Squares Df Mean Square F Sig.1 Regression 2.468 1 2.468 4.391 .037a



Residual 141.057 251 .562Total 143.525 252


The P- value of 0.37 implies that there is no linear regression relationship between

days of the month and the Daily return of stock index at 5 percent significance level.

Table 8: Year 2005 Model summaryModel

RR Square

Adjusted R Square



F Change df1 df2

Sig. F Change

dimension0 1 .015a .000 -.004 .54664 .000 .057 1 246 .811a. Predictors: (Constant), Days of the month

The coefficient of determination (R2) equals 0.000. This shows that days of the month do not explain any variation of Daily return of stock index but determined

by other factors. The P- value of 0.811 implies that the model of Daily return of stock index is not significant at the 5 percent significance level.

Table 9: ANOVAModel

Sum of Squares df Mean Square F Sig.1 Regression .017 1 .017 .057 .811a

Residual 73.507 246 .299Total 73.524 247






RR Square

Adjusted R Square



F Change df1 df2

Sig. F Change



The coefficient of determination (R2) equals 0.000. This shows that days of the month do not explain any variation of

Daily return of the stock index. The P- value of 0.999 implies that the model of



Daily return of stock index is insignificant at the 5 percent significance level.

Table 11: ANOVAModel


Residual 99330.000 244 407.090Total 99330.002 245




RR Square

Adjusted R Square



F Change df1 df2

Sig. F Change



The coefficient of determination (R2) equals 0.002. This shows that days of the month only explain 0.2% of the total variation of Daily return of the stock index

leaving 99.8% unexplained. The P- value of 0.529 implies that the model of Daily return of stock index is not significant at the 5 percent significance level.



Residual 254.763 246 1.036Total 255.176 247




days of the month and the Daily return of stock index at 5 percent significance level

.




RR Square

Adjusted R Square



F Change df1 df2

Sig. F Change

dimension0 1 .198a .039 .035 1.54444 .039 10.100 1 247 .002a. Predictors: (Constant), Days of the month


The coefficient of determination (R2) equals 0.039. This shows that days of the month only explain 3.9% of the total variation of Daily return of the stock index

leaving 96.1% unexplained. The P- value of 0.002 implies that the model of Daily return of stock index is not significant at the 5 percent significance level.


Sum of Squares df Mean Square F Sig.1 Regression 24.091 1 24.091 10.100 .002a

Residual 589.165 247 2.385Total 613.256 248






RR Square

Adjusted R Square



F Change df1 df2

Sig. F Change



The coefficient of determination (R2) equals 0.000. This shows that days of the month do not explain any variation of the Daily return of the stock index. The P-

value of 0.840 implies that the model of Daily return of stock index is significant at the 5 percent significance level.

Table 17: ANOVAANOVAb



ModelSum of Squares df Mean Square F Sig.

1 Regression .040 1 .040 .041 .840a

Residual 244.421 250 .978Total 244.461 251


Source: Research dataThe P- value of 0.840 implies that there is a linear regression relationship between



RR Square

Adjusted R Square



F Change df1 df2

Sig. F Change



The coefficient of determination (R2) equals 0.000. This shows that days of the month do not explain any variation of the Daily return of the stock index. The P-

value of 0.989 implies that the model of Daily return of stock index is significant at the 5 percent significance level.


Sum of Squares Df Mean Square F Sig.1 Regression .000 1 .000 .000 .989a

Residual 105.269 251 .419Total 105.269 252



The P- value of 0.989 implies that there is a linear regression relationship between



R R Square

Adjusted R

Std. Error of

Change StatisticsR F df1 df2 Sig. F



Square Change Change Change



The coefficient of determination (R2) equals 0.014. This shows that days of the month explain only 1.4 percent of total variation of the Daily return of the stock index leaving 98.6 percent to be explained

by other factors. The P- value of 0.063 implies that the model of Daily return of stock index is not significant at the 5 percent significance level.


Sum of Squares df Mean Square F Sig.1 Regression .021 1 .021 3.490 .063a

Residual 1.499 248 .006Total 1.521 249


Source: Research dataThe P- value of 0.063 implies that there no linear regression relationship between days

of the month and the Daily return of stock index at 5 percent significance level.

Conclusion The study did not find evidence to confirm that TOM effect exists at NSE. The study was conducted by dividing the month in to two parts, the TOM and the ROM. Daily returns for both TOM and ROM were computed from the daily indices of the NSE 20 share index. In all the years, TOM coefficients were insignificant as was the overall TOM coefficient as shown in table 1. Though TOM mean returns were higher than ROM returns in most of the years as shown in table 2, the regression analysis did not produce significant TOM coefficient to conclude that TOM effect exists at the NSE. In all the ANOVA tables above, p-values are less than 0.05 indicating that the TOM coefficient is insignificant. Besides the p-value being less than 0.05, looking at the correlation analysis in table 1, it can be noted that all

correlations for all the years are less than 0.7. This shows that the model is inadequate to explain the variations in securities returns. The existence of TOM effect at the NSE could have been an indicator that the capital market is not efficient. In conclusion therefore, it can be stated that no TOM effect was established at the NSE. This may be an indication that NSE efficiency is improving. However, other factors like the size of the index may have contributed to the results. The returns in TOM are slightly more than the return in ROM but not significant. This means that traders cannot post higher returns by trading during the TOM days. This is different from results of previous studies conducted in the developed world. Kenya being a developing country, may present a



scenario that may be the case for the developing nations security markets.

Implication on Policy and PracticeThe academia will find the study as a spring board for further studies. The results may be incorporated in future studies as reference work. Further, various studies have been done at the NSE but less on TOM. This study will therefore bridge this knowledge gap. The results of this study will be of benefit to the NSE and the CMA as a basis for policy formulation aimed at improving capital market efficiency. TOM as an anomaly to the EMH is not desirable if the market is efficient and so the need for market regulators to develop policies that improve market efficiency.Finally, the results of this study are likely to lend vital information to the brokerage firms as critical advisors to investors. They will be able to advise their clients on the best time of the month to sell or buy securities. Further, Investors shall find the results of the study useful as a guide on the best time to invest or divest their securities. The results of a TOM effect study shall appeal to institutional investors concerned with the timing of purchases and sales of securities in the stock market.

ReferencesAgrawal, A. & Tandon, K. (1994). Anomalies

or illusions? Evidence from stock markets in eighteen foreign countries. Journal of International Money and Finance 13, 83-106.

Ariel, R. (1987). Monthly effects in stock returns, Journal of Financial Economics, 18, 161- 174.

Bachelier, L. (1900). The random character of stock market prices. Cambridge, MA: MIT Press; 17 – 75, 1964.

Banz R. (1981). The relationship between return and market value of common stocks. Journal of Financial Economics 9, 3-18.

Cadsby, C., & Ratner, M. (1992). Turn-of-the-month and pre-holiday effects on stock returns: Some International evidence. Journal of Banking and Finance, 16, 487-510.

Dickinson and Marangu (1994). Market efficiencies in developing countries: A case of NSE. Journal of Business Accounting and Finance. January, 21(1), 133-150.

Dimson E. and Marsh P. (1986). Event study methodologies and the size effect. Journal of Financial Economics, 17, 113-142.

Fama E.F. (1970). Reply to efficient capital market: A review of theory and empirical work. Journal of Business Finance, 25, 383-417.

French, K. (1980). Stock returns and the weekend effect. Journal of Financial Economics, 8, 55-69.

Gibbons, R. and Hess P. (1981). Day of the week effects and assets returns. Journal of Business, 54, 4, 579-596.

Grossman, S.J. and Stiglitz, J.E. (1980). On the impossibility of informationally efficient markets. The American Economic Review, 70, 3.

Hanzel, R. and Ziemba,W.(1996). Investment results from exploiting TOM effects. Journal of Portfolio Management, 22, 17 – 23.



Kendall, M. G. (1953). The analysis of economic time series, Part I: Prices. Journal of Royal Statistical Society, 96, 1, 11-25.

Levin, R. and Zervos, S (1998). Stock markets, banks and economic growth. American Economic Review, 88, 537-58.

McKinley, T. (1999). Equity Returns at the Turn of the Month. Financial Analysts Journal, 64: 49-64.

Pettengill, G., & Jordan, B. (1988). A comprehensive examination of volume effects and seasonality of daily security returns. Journal of Financial Research, 11, 57-70.

Stiglitz, M. (1980) Securities market efficiency in an Arrow Debreu economy. The American Economic Review, 65(5), 812-24.

Zafar, R, Shah, S. and Urooj, F. (2009). Calendar Anomalies: Case of Karachi Stock Exchange. Research Journal of International Studies, 9, 88-99.

Manpower Development In The 21st Century: The Role Of Distance Education

Guantai Mboroki, PhD.1and Lydiah Wambugu, PhD.2

Generally, education is considered one of the biggest instruments for development; a means for realizing social, cultural, economic and political needs and aspirations. However, in many African educational institutions there is an enormous challenge in training a cadre of highly qualified professionals to fuel such development. There are inadequate educational resources, due to loss of the best talented faculty to the developed world. In addition, contemporary educational thought holds that one of the pivotal causes of inadequate school performance is the inability of schools to adequately staff classrooms with qualified teachers, especially in fields such as mathematics and science (Gordon and Thomas, 2007, p.43). To address the issue of teacher shortages in Kenya, some single mode universities converted themselves into dual-mode. Among 1 Senior Lecturer, School of Continuing and Distance Education, University of Nairobi, Nairobi, Kenya

[email protected] Lecturer, School of Continuing and Distance Education, University of Nairobi, Nairobi, Kenya.46 |



them is the University of Nairobi. Apparently, the belief among academics is that conventional education is real education. This makes distance education to act as complementary and worse of all supplementary. This paper sets out to compare whether there is a significant difference in TP performance between B.Ed (Science) Distance and on-campus students. A sample of 181 students: face-to-face n=131 vs. distance learning n= 50 students was used. The instrument of data collection was an observation guide. Though students taking courses by on-campus mode outperformed their counterparts in the distance mode of learning, this paper will conclude that distance learning should be treated as the emerging standard of quality in higher education and can effectively perform a complementary function, which should alleviate teacher shortage in Kenya. This study has demonstrated that science can be successfully taught by distance mode even given the current technological investment levels. This is important for policy because it leverages on what is available and the change that can bring especially in the area of manpower development, access and equity. The researchers argue that raising the technological level is not difficult since this can be done through collaboration with specialized bodies such as the African Virtual University (AVU) that is spearheading the infusion of ICTs and e-learning into the educational arena

Key words: Distance education, face-to-face learning, teaching practiceIntroductionConventional or on-campus mode of learning is the widely accepted and known mode of learning. Tsolakidis (2000)

defines conventional education/on-campus as the universally accepted approach for knowledge acquisition. On the other hand, distance learning is a

system of teaching by someone who is removed in space and time from the learner through teaching materials that have been systematically developed using different types of media to provide two way communication (Moore and Kearsley, 1996). By its inherent nature, distance education system has some features of

openness. Open learning describes a situation where control of learning is essentially in the hands of the learner. In literature, the terms Open and Distance Education (ODE) and Open and Distance Learning (ODL) are used synonymously and in some instances; this paper will treat the two terms as synonyms.

In Kenya, the Public Universities Inspection Board (2006) and Amutabi (2011, p.3) reports that in Kenya ODL is looked down upon and ‘very few universities have embraced the model of open and distance learning and even in those institutions where this is used, it is taken as a second or third class form of instruction’. This could be a pointer that the academics do not view the quality of distance learning as similar to

conventional learning. What is more, the academic performance of distance learners is deemed to be of lower quality. The Public Universities Inspection Board reported that:Experiences from other countries such as Tanzania, Nigeria, South Africa and the United Kingdom have demonstrated the potential of Open and Distance Learning (ODL) in increasing access. Kenya has not pursued this mode of delivery in a consistent and aggressive manner. As a


Z. B. Awino, 04/02/14,

Bold


result, Open and Distance Learning programs are the individual initiatives of the local universities with limited government funding (Republic of Kenya, 2006, p. 19).

Whereas the above concerns called for a national policy framework that would address the issue, this has not been done. Yet the number of students who desire university education keeps swelling every year. Some of these students are adults who cannot afford to enroll on a full-time

basis because of work, family responsibilities, and other commitments. Others are Kenya Certificate of Secondary Education (KCSE) candidates who attain the minimum university entry requirements of grade C+ as shown in Table 1.1 but miss admission into public universities because admission is tied on the bed capacity and resources allocated to each university by the government (University of Nairobi, 2008, p. 9). The present public universities in Kenya absorb a total of 28-30 percent (MoHEST, 2011).

Table 1.1: Number of KCSE candidates admitted in Private and Public universitiesYear C+ and Above Private universities Public universities Left out2006/2007 62,926 17,706 10,000 35,2202007/2008 82,134 20,432 16,000 45,7022008/2009 72,500 22,123 20,000 30,3772009/2010 81,048 35,179 24,221 21,6482010/2011 97,134 37,848 29,237 30,049

Source: The Kenya National Examination Council and Joint Admission Board (2011)

Those who are shut out may join private universities. Private universities absorb less than 25 percent of students who attain mean grade of C+ and above at KCSE level (MoHEST, 2011). Others may join the public universities through the ‘parallel programs’ commonly referred to as Module 2 and Module 3 programs. In Module 2 students learn in the evenings, or get integrated with the fulltime students while Module 3 is the distance learning mode.

While ODE is credited for enhancing participation and access in higher education, there is skepticism on the quality of learning. This is on the assumption that it undermines traditional education, limit student interaction with peers and lecturers and eradicate the platform for which a deliberate academic discourse takes place (Mathews, 1999). Other scholars assert that distance education has shown a capacity to help students acquire knowledge and communication skills but fails in developing the skills of analysis, synthesis,



application, judgment and value that are the ‘characteristics of a truly higher education’ (Smith and Kelly, 1987, p. 48).

Fox (1998) argues that what is in dispute is not whether distance education is ideal, but whether it is good enough to merit a university degree, and whether it is better than receiving no education at all. Fox alludes to an argument that students learn far too little when the teachers personal presence is not available because the student has more to learn from the teacher than the texts. Hannay and Newvine (2006, p. 5) add that distance learning examinations are ‘cheap’ since they test lower cognitive levels of knowledge and comprehension. No wonder, distance education has been described as learning at the back door (Ding, 1988); second hand education or the Cinderella of the educational spectrum (Keegan, 1990) and an ignored and neglected step child of education (Peters, 1993).

Despite such concerns, the researchers are of the view that ODL is as good as on-campus learning as long as the students are stimulated to order their thoughts and actively get involved in the academic requirements. This is because there are successful single and dual mode universities in the world that have continually channelled out quality graduates into the labour market. For instance in 2011, the Open University of the United Kingdom was rated among the United Kingdom's top three higher education institutions. It came third out of 157 universities in the National Student Survey - placed equally with Oxford. The rating was based on student satisfaction, number of graduates and research output (http//www.open.ac.uk/about/main). Open

University of Tanzania and University of South Africa are among the top ranked universities in Tanzania and South Africa respectively with various programs on offer. (http//www.open.ac.uk/about/main)

Assessment of Academic Performance of B.Ed (Science) CourseAssessment may be defined as the process of determining the extent to which learning has taken place. Chris (1999) defines assessment as the systematic collection of information about student learning using the time, knowledge and resources available, in order to inform decisions about how to improve learning. In this study, assessment is defined as the methods used by lecturers to determine the extent to which students have comprehended the course content.

University of Nairobi uses two methods to assess the Bachelor of Education (Science) students: written examinations at the end of each semester and teaching practice at the end of the third year. Examinations consist of 30% course work comprising of written assignments (term papers), semester tests, practicals, projects; and 70% written examination at the end of each semester. Each course unit is examined by a two hour paper and the pass mark for each paper is 40%. The B.Ed (Science) students in the two modes of study sit for different but equivalent examinations which go through the same university mechanism of content validity through internal and external moderation. Teaching is a complex activity; student teachers need to develop their capacity in making intelligent decisions to handle ambiguous and challenging situations in schools and beyond. Lam and Fung (2001, p. 1) notes that teacher education ‘…is



largely a matter of developing a teacher’s capacities for situational understandings as a basis for wise judgment and intelligent decisions in complex, ambiguous and dynamic educational situations’. This means that educational and teaching theories learned in teacher education programs may not be applied straightforwardly to complex and dynamic educational situations. It is only during teaching practice when student teachers are given an opportunity to test theories learnt in the lecture rooms in real life experiences. No wonder teaching practice is compulsory to all students studying for a degree in Education.

Teaching practice is equivalent to two units. Each candidate on teaching practice is assessed a minimum of three times in each of the teaching subjects and each assessment is marked out of 100 marks. Teaching Practice is viewed by many educational authorities as the measure for the success of an Education degree (Stones 1992)… ‘because the student teacher is provided with the opportunity of practically synthesizing and applying in a real situation the theoretical learning that has been provided throughout the teacher preparation program’ (Molomo as cited in Digolo, 2002, p. 96). Teaching practice students are assigned students to teach and what they teach is real learning content that might decide the future of that learner. At the University of Nairobi, students go for their teaching practice just before the beginning of their final year. By this time they have covered most of their academic work and they have covered both general and subject methods of teaching. Therefore, it is not enough for a person to show examination grades, but should demonstrate that learning through

communicative acts (Keegan, 1990, pp. 14-15). Mager (1973, p.2) aptly concludes that ‘you cannot find out if someone can ride a unicycle unless you or someone else watches him ride one. You cannot find out if the objective is achieved unless you use items that ask the student to perform what the objective is about’ (Mager, 1973, pp. 2-3). This then implies that you cannot find out if a student-teacher can translate theory into practice unless you watch him or her in a real classroom environment.

Literature ReviewThere are numerous research studies that have been conducted comparing the two modes of learning. These studies have presented conflicting conclusions creating ‘a doubt, a barrier, and an indeterminate situation crying out to be made determinate’ (Kerlinger and Lee, 2000, p. 67). This will be demonstrated below:

Comparative Studies in KenyaMutonga (2011) conducted a comparative study in Kenya of academic performance of students in the Registered Community Health Nurse Upgrading Program under face-to-face and Distance learning mode of instructional delivery. The design of this study was descriptive survey. The target population was 1,363 students; 943 distance study and 420 face-to-face students who sat for the Nursing Council of Kenya (NCK) licensing examination between 2008 and 2010. The study used secondary data obtained from NCK’s electronic database. The study found out that there was a statistically significant difference (p=0.000; p< 0.05) between the performance of distance education and face-to-face students. Face-to-face students performed significantly higher than distance study students. The study



also found that there was a relationship between student’s performance and their entry qualification. The higher the O-level grade attained, the higher the licensing examination mean score.

Mboroki (2007) carried out descriptive Survey study that compared performance in teaching practice between 43 Bachelor of Education (Arts) on- campus students and 50 distance study students of the University of Nairobi. The findings of this study showed that there is parity in performance in teaching practice between on-campus and distance study students of the University of Nairobi in the way the students are assessed in lesson preparation, presentation, development and mastery of content, classroom management and personal deportment.

Comparative Studies outside KenyaUnterberg (2003) carried out a study at the Harvard Medical School Center for Palliative Care in Boston. The purpose of the study was to compare learning outcomes of undergraduate students in a Physical Therapy Education degree course, given different learning environments. The two learning environments were in-classroom environment, where students met in a classroom with the instructor present, and the computer-mediated distance learning environment, where students worked independently, receiving information and communicating with the instructor via e-mail. Unterberg’s study found out that 43.0 percent of the distance group and 49.0 percent of the in-class group scored above-average, 32.0 percent and 34.0 percent scored average while 25.0 percent and 17 percent scored below average respectively.

Urtel (2008) carried out a study at the Mid-Western Urban University, Indianapolis to compare academic performance between face-to-face and distance education course format. The sample was 116 face-to-face and 269 distance study learners. Data were collected by use of document analysis. Examination grades and gender were used as comparative variables. Results analyzed at .05 level of significance indicated that there was a statistically significant difference between the distance education and face-to-face groups regarding overall academic performance as measured by grade earned (p=0.011). The students enrolled in the face-to-face mode earned, overall, an average GPA 3.16/4.00 against 2.28/4.00 of the distance study students. When analyzing gender interactions, there was no statistically significant difference in performance of females versus males (p=0.2214). Urtel concluded that contrary to some studies, students in distance education course do not automatically perform equally as well, or even better, than in a face-to-face course and that older students’ in distance education do not automatically outperform their younger counterparts in face-to-face class.

Lizzio, Wilson and Simons (2002) conducted a study to determine the influence of Tertiary Entrance (TE) score in the students’ years 11 and 12 of their secondary education on academic outcomes at the University of Griffith, Australia. A sample of 64 students was drawn from the Faculty of Business Studies. Findings showed that TE score was positively but weakly (ρ = .39) associated with a high Grade Point Average (GPA) score measured on a scale from 1 (low) to 7 (high). The weak



relationship could be a pointer that past performance is not a determinant of the present performance probably due to maturity which is associated with cognitive development. In contrast, Adedeji (2001) found a strong positive correlation (ρ = .85) between students admission scores and their undergraduate performance at the Faculty of Technology, University of Ibadan. A study by Aderson, Benjamin and Fuss (1994) also found out that student’s who received better scores (between 777-999) in high school tended to have a Grade Point Average score of above 3.0/4.0 at the university.

Pascallera and Collins (2003) conducted a randomized instructional experiment between a group of 46 students in a Fire Fighting Tactics and Strategy course learning on campus and learning at a distance at a Community College in Iowa. In one format students received face to face instruction in a traditional classroom on-campus while the rest received the same course instruction by two-way Interactive Television on the Iowa Communication Network (ICN). The findings of the study were that: The specific medium of instruction has little impact on how students learn. Students can master course facts and concepts as well when they receive instruction at a distance as they can when they receive the same instruction on-campus in a traditional face to face environment; and the parity in outcomes holds irrespective of students pre-course level of content knowledge, their prior exposure to post-secondary education or their related professional training and experience. Pascarella and Collins’ study used a different form of distance education; teleconferencing (e-learning) which is more interactive

(Keegan, 1986) than the print medium used at the University of Nairobi.

Carmel and Gold (2007) examined the relationship between modality of course delivery and the level of student Grade Points Average (GPA), satisfaction and retention achieved by students attending either traditional on-site or distance study courses at the University of Phoenix. A document analysis incorporated data from 110 courses and 164 students of which 95 students were attending on-site classes while 69 were in distance study. The average GPA for distance mode was 3.74 while on-site was 3.77. Statistical significance of the means was tested at α = 0.05. Carmel and Gold concluded that there is no statistically significant difference between the means of the groups (p=0.293).

Cano and Garton (1994) conducted a study that investigated in-service performance of 82 practicing teachers who had majored in Agricultural Education at The Ohio State University. These researchers were interested in the relationship between instructional format and student’s success in a college level course. The researchers based their study on Richard Clark’s theory which states that any medium of instruction is capable of delivering instruction. Cano and Garton (1994). Two instructional formats were compared; face-to-face learning and distance learning with minimal face to face interaction. Results indicated a low positive relationship (ρ = .30) between instructional format and academic achievement in the course. The researchers concluded that instructional medium does not significantly influence academic performance.

52 |



Neuhauser (2002) conducted a research that focused on the relationship between learning environment and academic outcomes. The researcher analyzed documents of 54 students enrolled in two high school psychology classes. 27 students were in distance mode and the other 27 were in the face-to-face mode of learning. Results showed that the mean performance of the face-to- face group was not significantly higher (M =76.47) as compared to distance mode mean performance of 74.21 with p>.05. Thus the researcher concluded that equivalent learning activities could be equally effective for distance and traditional classroom learners as long as the students spend ‘quality time’ studying.

Okoh (2010) examined the influence of age, financial status and gender on academic performance among a sample of 175 undergraduates; 57 males and 118 females enrolled in a Counseling Psychology course offered through face-to-face and distance learning. One of the hypotheses the researcher tested was ‘There is no significant difference in the academic performance of undergraduate students based on gender’. A p> 0.05 was obtained and the researcher concluded that there is no significant difference between male and female respondents in their academic performance. A probable explanation to this observation is that since both male and female undergraduates are exposed to same social and academic environment, they are assessed using the same criteria, then their reaction to grading (academic performance) is similar.

Magagula and Ngwenya (2004) carried out a comparative analysis of the academic performance of off-campus (distance) and

on-campus learners at the University of Swaziland in 2004. The purpose of the study was to; a) examine the background characteristics of 70 off-campus and 70 on-campus enrolled in Bachelor of Arts and b) the extent to which the academic performance of off-campus and on-campus were similar and/or different. Academic performance was operationalized as the overall average mark or grade obtained by a learner in each of the following subjects in year one of the final examination at the University of Swaziland; Academic Communication Skills (ACS), History, Theology, African Languages, Geography, and English. The researchers compared the academic performance based on Rumbles (1997) assertion that if the entrance requirements (entry qualification) and content of both off-campus and on-campus programs are the same, lecturers are the same and both off-campus and on-campus learners write the same final examination then it should be possible to compare the academic performance of both modes.

Research MethodsThe design of this paper was a survey where a sample of 50 ODE and 131 on-campus students was drawn from a population of 58 ODE and 195 on-campus students respectively. These were the third year B.Ed (Science) students who were out for TP in 2012. Simple random sampling was used. To achieve a simple random number, each student in the sampling frame was assigned a number from 1 to 195 for on-campus students and 1-57 for distance study students. Computer Random Number Generator was then used to generate a list of random numbers that were used to the sample.



An observation guide was used to measure teaching practice performance. The university uses a standardized measuring instrument referred as Teaching Practice Observation Sheet for awarding marks. Various behaviors (or elements) are assessed. These behaviors are Preparation (20 marks), Presentation (15 marks), Lesson development (35 marks), Classroom management (10 marks), Summary/conclusion (10 marks) and Personal factors (10 marks). Results and DiscussionThe total student sample size for this study was 181 B.Ed (Science) students. 131 representing 72.4% were on-campus students while 50 representing 27.6% were

distance study students. Majority of the on-campus students were males representing 71.0% while majority of the ODE students were females representing 56.0% as shown in Table 1.2. This high percentage of the males could be a pointer that there is a gender gap in science performance at KCSE. As per this finding, males had performed higher in Sciences at KCSE than females. The ODE mode of learning attracted more females than males a finding that concurs with Aragon et al (2002) study who found that females enrolled in distance education classes at a higher rate than males because of the flexibility of the mode of learning.

Table 1.2: Gender Distribution of the B.Ed (Science) students

Distance Study students On-campus students

Gender Frequency Percentage Frequency Percentage

Male 22 44.0 93 71.0Female 28 56.0 38 29.0TOTAL 50 100.0 131 100.0

Teaching Practice (TP) is a compulsory exercise for every student enrolled in a B.Ed (Science) course. A common adapted T.P Lesson Observation Guide was used to assess the students as they taught in their TP schools. Just like in examinations, the TP marks are converted into grade letter where Grade A ranges from 70-100%, Grade B from 60-69%, Grade C from 50-59%, Grade D from 40-49% while E (fail) is from 0-39%.

The researchers sought to find out teaching practice mean performance between ODE and on-campus students. First, the researcher determined the teaching practice mean score performance translated into letter grades. As shown in Table 1.3, majority of the students in both modes scored above 60.0% while none scored below 49.0%. For instance, 27 (54.0%) ODE and 92 (70.2%) on-campus students scored Grade A in TP.



Table 1.3: Letter Grade Performance of the B.Ed (Science) students in teaching PracticeGrade performance Distance study students On-campus students

Frequency Percentage Frequency PercentageA (70-100%) 27 54.0 92 70.2B (60-69%) 21 42.0 38 29.0C (50-59%) 2 4.0 1 0.8D (40-49%) 0 0.0 0 0.0TOTAL 50 100.0 131 100.0

As per the university evaluation, this performance may be termed as good. This may be attributed to the preparation and support the students received from the university and also from the regular teachers in their TP schools. Using the scores assigned to the student, the overall TP mean score performance for the on-campus students was higher than for the ODE students. As shown in Table 1.4 the

mean score performance for the on-campus students was 71.73 against a mean of 69.88 for the ODE students with a standard deviation of 4.59 and 4.77 and SE of .40 and .67 respectively. Based on the normal curve, approximately 68% of the on-campus students scored between 67.14 and 76.32 while ODE students scored between 65.11 and 74.65.

Table 1.4: Mean score performance in TP

TP Performance Distance study students On-campus studentsMean N Standard

deviationMean N Standard

deviationOverall performance in TP

69.88 50 4.77 71.73 131 4.59

The relationship between the mode of learning and the overall TP performance shows a weak positive linear relationship, Pearson correlation coefficient of r = .176 as presented in Table 1.5. The Table also shows that this relationship is significant p < .05. This finding implies that there is a

significant relationship between the mode of learning and TP performance. On-campus students perform significantly higher in TP than the ODE students despite the latter having prior Training in Education.

Table 1.5: Correlation between TP performance and mode of learningOverall Teaching practice Performance

Mode of Learning

Overall Teaching practice Performance

Pearson correlationSig. (2-tailed)

1.181

.176*

.018 181



NMode of Learning Pearson

correlationSig. (2-tailed)N

.176*

.018 181

1.181

*Correlation is significant at the 0.05 level (2 tailed)

Finally, the researchers sought to determine whether the difference in mean performance between ODE and on-campus is significant. Independent t-test was used and the output is presented in Table1.6. As shown in the table, there is a significant difference between the mean performance of ODE students and their counterparts in the on-campus mode of learning. The students in the on-campus mode of learning performed significantly higher (M=71.33, SE =0.401) than students in the ODE mode of learning (M=69.88, SE=0.65). This difference was found to be significant t (179) = -2.392, p<.05.

The behaviors assessed during teaching practice were categorized into six attributes. These are lesson preparation, lesson presentation, lesson development, classroom management, summary/conclusion and personal factors. There is no significant difference in the performance of students in both modes of learning in Preparation of the lesson (p=.618), Presentation of the lesson (p=.508), Classroom management (p=.141) and summary and conclusions (p=606). However, on-campus performed significantly higher in lesson development (p=.000) while ODE performed significantly higher in personal factors (p=.036.

Table 1.6: Output for the Mean performance in each attributeTeaching practice attributes

Mode of learning

N Mean Std. deviation

Std. Error Mean

t df Sig. (2-tailed)

Overall TP mean performance

Distance learningOn-campus learning

50

131

69.88

71.73

4.77

4.60

.675

.401

-2.392 179 .018

Preparation Distance learningOn-campus learning

50

131

11.54

11.37

1.88

2.17

.265

.189

.499 179 .618

Presentation Distance learningOn-campus learning

50

131

10.74

10.60

1.12

1.37

.159

.120

.664 179 .508

Lesson Development Distance learning

50 26.16 3.06 .433 -5.128 179 .000



On-campus learning

131 28.34 2.35 .205

Classroom management


50

131

6.72

6.96

.97

.99

.137

.086

-1.480 179 .141

Summary and conclusion


50

131

6.60

6.69

.78

1.20

.111

.105

-.519 179 .606

Personal factors Distance learningOn-campus learning

50

131

8.10

7.82

.93

.75

.132

.066

2.115 179 .036

The researchers expected the ODE students to perform exceptionally higher than the on-campus owing to the fact that majority of the ODE students (92.0%) as shown in Table 4.8 have prior training in Education. On the contrary, the on-campus students performed significantly higher than the ODE. This might seem as a challenge to the finding by Mboroki (2007) that there is parity in teaching practice performance between on-campus and distance study students. His findings were on a bachelor of Education (Arts) program and may differ from a science program. An arguable point is that the University of Nairobi may be assuming that the ODE students, majority of whom are practicing teachers do not need to go through micro teaching while at the university. This therefore means that the student goes to the field without being exposed to a mock secondary school classroom. This finding is in accord with Kumar (1997) assertion that conventional education leads to the development of oral presentation skills and interpersonal skills as a result of high teacher/learner and

learner/learner interaction. They may therefore not have the skills to teach in a secondary school. As Odumbe and Kamau (1986) explain, ODE students lack confidence in themselves because the society favors conventional modes of teaching.

ConclusionsIt is the assertion of the researchers that lack of interactivity which is presumed to be a major contributor to the difference in performance between the two programs will be overcome when the University of Nairobi fully integrates the e-learning component that is being provided by the partnership with the African Virtual University.

Implication on Policy and PracticeIt is not clear why Kenya has so far not developed a policy framework that would mainstream open distance education into the formal delivery of education. Kenya has a long history of distance education experience spanning from 1967 when the Correspondence Course Unit was



established at the University of Nairobi’s Kikuyu Campus and shortly followed by the very successful In-service Teacher Training Programmme of the Ministry of Education in 1969 at the same venue. Indeed so successful have been the Kenyan experiences that countries like Uganda, Tanzania, Zimbabwe, Botswana, Swaziland, Namibia, Ghana, Nigeria and even India have borrowed the Kenyan model and introduced open universities and colleges.The University of Nairobi has been running a Bachelor of Education (Arts) programme since 1986 and the science equivalent since 2003. The results of this study have demonstrated that a science programme can be taught successfully at a distance. 99.2% of the distance study students scored over 60% in an examination where a 40% mark would have been adequate for a pass. The fact that the research compared both on-campus and distance education students using the same examination dispenses with the notion espoused by Hannay& Newvine (2006) that distance education examinations are ‘cheap’ since they test lower cognitive levels of knowledge and comprehension. It confirms the assertion by Fox(1998) that learning by distance education mode is good enough to merit a university degree.The study further confirms that distance education brings more female learners to the learning arena at the tertiary level. This is an encouraging finding because it demonstrates a direction that can foster gender equity through education. What the findings of this study mean for policy is that distance education can be used to break the bottleneck of qualified manpower in Kenya. Research has shown that the Gross Enrollment Ratio for Kenya

has been 3% against Africa’s 6%(Kenet 2009). Some educational economists have observed that at least 12-15% of a nation’s workforce must have tertiary education if it is going to compete in the new global economy. They also add that ‘Seeking to meet this demand requires a conceptualization of massification that is not currently under consideration’ (Taferra and Altbach, 2003:74). In recent years the Government has come out strongly in rhetoric in support of distance education as the option that will bring about the massification that will break the manpower bottleneck. There are plans to establish an open university by the end of this year(2014). Among the policy documents that have called for the establishment of the Open University include the Sessional Paper No. 1 of 2005 on Education, Training and Research (2005), Report of the Public Universities Inspection Board (2007), National Strategy for University Education (2008) and the Road Map for Open University (Rumble Report 2008). Others include the Task Force on the Realignment of the Education Sector to the 2010 Constitution of Kenya (2012), the Task Force on Alignment of Higher Education, Science and Technology sector to the 2010 Constitution of Kenya (2012), the Sessional Paper No. 14 of 2012, on reforming education and training sector in Kenya and the Universities Act, 2012,

This study goes some way in reaffirming the capacity of distance education to produce graduates who are as competent as those produced by the on-campus system.

ReferencesAdedeji, O. B. (2001). A study of the

Relationship between Students UME



Results and their Undergraduate Performance, (unpublished Master’s Thesis). University of Ibadan, Nigeria

Aderson, G., Benjamin, D. & Fuss, M. (1994). The Determinant of Success in University Introductory Economics Courses. Journal of Economic Education, Vol. 25, No,2, pp. 99-120

Amutabi, M. N. (2011). Making a Case for Open and Distance Learning in Kenya: Possibilities and Prospects. A Paper Presented at the Distance Education Association of Southern Africa (DEASA) Conference on 30th September – 2nd October 2011, Dar es Salaam

Aragon, S. R., Johnson, S. D., & Shaik, N. (2002). The influence of learning style preferences on student success in online versus face-to-face environments. The American Journal of Distance Education, 16 (3), pp.227-244.

Cano, J., & Garton, B. L. (1994). The Relationship between Agriculture Pre-service Teachers Learning Styles and Performance. .Journal of Agricultural Education, Vol 35, No. 2, pp. 6- 10.

Carmel, A., & Gold, S.S. (2007). The effects of Course Delivery Modality on Student Satisfaction and Retention and GPA in On-site vs. Distance Courses. Turkish Online Journal of Distance Education, April 2007, ISSN 1302-6488, Vol. 8, No. 2, Article 11

Chris, M. (1999). Selecting Methods of Assessment, Unpublished material for Southern Cross University booklet. Retrieved from http:/www.brookes.ac.uk/assessment

Digolo, P. (2002). Visions on Teacher Education in Kenya: A Teaching Practice Perspective. The Fountain: Journal of the Faculty of Education, Vol. 1, No. 1, pp. 90-102

Ding, X. (1988). An Introduction to Distance Higher Education. Beijing: CRTVU Press

Fox, J. (1998). Distance education: Is it good enough? The University Concourse. Retrieved from http://www.clearspoint.com/accelepoint/articles

Gordon M and O’Brien Thomas V(ed). (2007), Bridging theory and practice in teacher education: The Netherlands, Sense publishers

Hannay, M. & Newvine, T. (2006). Perceptions of Distance Learning. A Comparison of Online and Traditional Learning. Journal of Online Learning and Teaching, Vol. 2, No. 1, pp. 19-29

Keegan, D. (1990). Foundations of Distance Education (2nd ed). London: Routledge

KENET (2009). E-readiness survey of East African Universities Report 2009. Nairobi: Kenya Education Network.

Kerlinger, F., & Lee, H. (2000). Foundations of Behavioral Research. New York: Harcourt College Publishers

Kumar, K. (1997). The Need for Place. In Smith, A., & Webster, F. (Eds.), The

Postmodern University: Contested Visions of Higher Education in Society. Buckingham: Open University Press

Lam, J. & Fung., Yan Mok. (2001). Strengthening Teacher Training Programs: Revamping the Model of Teaching Practice. A Paper Presented at the 2nd Hong Kong Conference on Quality in Teaching and Learning in Higher Education on 3rd – 5th May 2001

Lizzio, A., Wilson K., & Simons, R. (2002). University Students Perceptions of the


http://www.clearspoint.com/accelepoint/articles

http://www.clearspoint.com/accelepoint/articles


Learning Environment and Academic Outcomes: Implications for Theory and Practice. Studies in Higher Education, Vol. 27, No. 1, pp 27-52

Magagula, C.M. & Ngwenya, A.P. (2004). A Comparative Analysis of the Academic Performance of Distance and On-campus learners. Turkish Online Journal of Distance Education, Vol. 5, No. 4, pp 17-26

Mager, R.F. (1973). Measuring Instructional Intent, California: Fearon Pitman

Mathews, D. (1999). The Origins of Distance Education and its Use in the United States. The Journal 27, Vol.6, No. 2, pp. 54-56

Mboroki J.G. (2007). A comparative study of performance in Teaching practice between the Bachelor of Education (Arts) on-campus and Distance study students: A case study of University of Nairobi. (Unpublished PhD Thesis). Moi University, Kenya

Ministry of Higher Education, Science and Technology. (2011). MoHSET News. Kenya:Government Printers

Moore, M. G., & Kearsley, G., (1996). Distance Education: A Systems View. Belmont: Wadsworth

Mutonga, J. W. (2011). A Comparative Study of Student Academic Performance under Face-to-face and Disance Learning Mode of Instructional Delivery: A Case of the Registered Community Health Nurse Upgrading Program, Kenya. (Unpublished Master’s Project). University of Nairobi, Kenya

Neuhauser, C. (2002). Learning style and effectiveness of distance and face-to-face Instruction. The American Journal of Distance Education, 16, 99-113.

Odumbe J and Kamau, J. (1986). Student’s Handbook. Faculty of External Studies. University of Nairobi: Nairobi

Okoh-Ebenuwa, E.E., (2010). Influence of Age, Financial Status and Gender on Academic Performance among Undergraduates. Journal of Psychology, Vol. 1, No. 2, pp. 99-103

Pascarella, E., & Collins, J. (2003). Learning on Campus and Learning at a Distance: A Randomized Instructional Experiment. Research in Higher Education, Vol. 44, No. 3 pp. 315-326

Peters O. (1993). Distance education in a Postindustrial Society. In Keegan, D. (Ed.), Theoretical Principles of distance education. New York: Routledge, pp. 39-58

Republic of Kenya. (2006). Transformation of Higher Education and Training in Kenya to Secure Kenya’s Development in the Knowledge Economy: Report of the Public Universities Inspection Board. Kenya: Government Printers

Smith, P. & Kelly, M. (1987). Distance Education and the Mainstream. Sydney:

Croom Helm Stones, E. (1992). Quality Teaching: A sample of cases. London: Routledge

The Open University. (2011). About the Open University. Retrieved from http//www.open.ac.uk/about/main

Teferra, D and Altibach P. (2003). African Higher Education. Indiana University Press: Bloomington and Indianapolis.

Tsolakidis, C. (2000). Distance Education: A second best in learning? Retrieved from url:http://toide.anadolu.edu


url:http://toide.anadolu.edu


University of Nairobi. (2008). Information Booklet. Nairobi: UON

Unterberg, M. (2003). Do Learning Style and Learning Environment Affect the LearningOutcome? Journal of Physical therapy education (St. Louis, Maryville University), Vol. 3(8), pp. 67-75

Urtel, M. G. (2008). Assessing Academic Performance between Traditional and DistanceEducation Course Formats. Educational Technology and Research Journal, Vol. II, No.1, pp. 322-330

Foreign Direct Investment And Economic Growth: An Empirical Analysis Of Kenyan Data

Daniel O Abala, PhD.1

The paper investigates the main drivers of real Gross Domestic Product growth in Kenya as well as those that drive the foreign direct investment (FDI) in Kenya. It is widely acknowledged that FDI has potential benefits that accrue to host countries. The view suggests that FDI is important for economic growth as it provides much needed capital, increases competition in host countries and helps local firms to become more productive by adopting more efficient technology. Kenya’s record in attracting FDI from the 1980s has been poor though it was a favoured destination in the 1970s.The study findings show that FDIs in Kenya are mainly market-seeking and these require growing GDPs, political stability and good infrastructure, market size as well as reduction in corruption levels. The prevalence of crime and insecurity would be impediments to FDI inflow. The policy implications of this study are that Kenya’s FDI’s tend to be mainly market seeking and for this reason policy makers in Kenya should focus on improving political stability, emphasize the development of good infrastructure and growing the country’s GDP. This should be coupled with a serious attempt at reducing corruption levels as well as a serious assault on the prevalence of crime and insecurity which are major impediments to this type of FDI inflows.

Keywords: Foreign Direct Investment, Economic Growth, Determinants, Kenya

1 Senior lecturer, School of Economics, University of Nairobi, Nairobi, Kenya [email protected] |



IntroductionForeign direct investment (FDI) in Kenya is defined as investment in foreign assets, such as foreign currency, credits, rights, benefits or property, undertaken by a foreign national (a non-Kenyan citizen) for the purposes of production of goods and services which are to be sold either domestically or exported overseas (Investment Promotion Centre Act, Chapter 518). FDI generally refers to an investment made to acquire a lasting management interest (normally 10% of voting stock) in a business enterprise in a country other than that of the investor defined according to residency (World Bank, 1996). Ownership of less than 10% is regarded as portfolio investment. Foreign direct investment has grown enormously in the last three decades. For example prior to the recent economic crisis, global FDI has risen to US $ 1,833 billion in 2007 well above the US $ 1,748 billion in 2000(UNCTAD, 2008). The production of goods and services by multinational corporations and their foreign affiliates have continued to rise as evidenced by increase in FDI from US $ 15 trillion in 2007 US $ 18 trillion in 2010 (UNCTAD 2010). The increase in FDI has been singled out as the most important factor for poverty reduction (Rose and Mwega, 2006). Most developing countries such as Kenya are interested in FDI a source of capital for industrialisation. This is because FDI involves a long term commitment to the host country and contributes significantly to the gross fixed capital formation.FDI has been identified to contribute significantly to the economic growth of countries. Governments of many host

countries (recipients of FDI) are using financial incentives such as tax allowances and grants in aid among other policies to attract FDI into their economies due to the perceived benefits associated with FDI inflows. It has been suggested in numerous papers that foreign firms are able to positively affect the levels of productivity and growth rates in the industries they enter and to also promote skill upgrading, increase employment and increased innovation (Blomström, 1986; Blomström and Persson, 1983; Görg and Strobl, 2001; UNCTAD, 2005). However, it has also been argued that FDI may lower or replace domestic savings and investment, transfer low level or inappropriate technologies for the host country’s factor proportions target primarily the host country’s domestic market and even inhibit the expansion of indigenous firms thereby limiting growth. By focusing solely on local cheap labour and raw materials, foreign firms may not be helpful in developing the host country’s dynamic comparative advantages (UNCTAD, 2005). Nevertheless, the negative consequences of FDI can be managed with proper business and labour regulation (Rose and Mwega, 2006; Kinuthia 2010).There are at least three major types of FDIs. The market-seeking FDI usually serves local and regional market and involves the replication of production facilities in the host countries. A variant of this type of FDI is also known as Tariff-jumping or export-substitution FDI and it is driven mainly by market size and market growth of the host economy. Due to market and income considerations FDIs in small and poor countries are unlikely to be



of the market seeking type (see Lim 2001; Campos and Kinoshita, 2003 and UNCTAD, 1998).The resource or asset-seeking FDI is another type of FDI and involves the relocation of parts of the production chain to the host country. This is usually driven by the availability of low-cost labour and is often export-oriented. This type of FDI is also attracted to countries with abundant natural resources such as oil and gas.The third type of FDI is the efficiency-seeking type where the firms gain from common governance of geographically dispersed activities in the presence of economies of scale and scope. The idea here is to take advantage of special features such as labour costs, skills of the labour force and quality of infrastructure. We next examine the evolution of FDI flows in Kenya and how it has affected economic growth in Kenya.The hope of vision 2030 has apparently not been fulfilled and Kenya’s share in the regional market, both in EAC and the wider COMESA is still less than 15%. However, it still appears that the economic growth of a developing country may well depend on among other things on an opportunity to make profitable investments and accumulate capital. It is similarly true that one of the ways of achieving this objective is through the attraction of foreign capital and investments which allows a country to exploit opportunities that would otherwise not be available (OECD, 2002).

Evolution of FDI and Kenya’s Economic GrowthKenya has had a long history with foreign firms. From independence of 1963 through the 1970s and part of the 1980s it was one of the most favoured destinations of FDI in

the Eastern Africa. FDI grew steadily through the 1970s as Kenya was the prime choice for foreign investors seeking to establish a presence in Eastern and Southern Africa. In the 1970’s Kenya was the most favoured destination for FDI in East Africa. However, over the years she has lost her appeal to foreign investors a situation that has continued to the present. In 2008, Kenya launched vision 2030 with the objective of among other things to achieve global competitiveness for FDI and gain economic prosperity. This initiative has seen a renewed commitment to attract FDI to assist in achieving higher economic growth rates. Kenya has had inconsistent trends of FDI inflows starting with the 1970-1980 period .The then relatively high level of development, good infrastructure, market size, growth and openness to FDI at a time when other countries in the region had relatively closed regimes all contributed to the multinational companies (MNCs) choosing Kenya as their regional hub. There was also relative political stability and security during the period. FDI started at a low of around US$ 10 million a year in the early 1970s before peaking at US$ 60 million by 1979-80. The country received relatively large capital inflows partly driven by rapid expansion in the agricultural sector, expansionary fiscal and monetary policies, sustainable budget deficit and the import substitution industrialisation (ISI) strategy. This involved overvalued exchange rates, import tariffs, quantitative restrictions and import licensing (Ikiara et al, 2003). Other factors included large and favourable regional markets from the original East African community (EAC) which attracted FDI into the country (World Bank, 2010). However, after the 1980s, Kenya’s economy was characterized by



deterioration in economic performance, corruption and bad governance. Inconsistency in the implementation of economic policies and structural reform measures as well as the deterioration of public service and infrastructure ensured decades of low level of FDI inflows. FDI inflows in the period 1981-1999 averaged only US$ 22 million per annum. It is noted that although Kenya was the leading destination of FDI in the East African region in the 1970s and 1980s the relative level of flows was never high even by developing countries’ standards. This can be seem by looking at the stock of FDI which was only 7.5% of the GDP in 2003, compared to 25.3% for Africa as a whole and 31.5% for developing countries (UNCTAD, 2005). Kenya’s regional leadership in attracting FDI also disappeared as soon as Tanzania and Uganda started reforming their economies and opening up to foreign investors in the early 1990s. FDI flows in the 1996-2003 period averaged some US$ 29 million annually while flows to Tanzania and Uganda surged to US$ 280 million and US$ 220 million respectively from negligible levels in the 1980s (see UNCTAD, 2005). In relative terms, Kenya’s case was even worse since its economy was about 30% larger than Tanzania’s and twice as big as Uganda’s in 2002. It is notable that developing countries as a whole attracted an annual average of US$ 41 of FDI per capita in 1996-2003 when Kenya only managed inflows of US$ 1.3 per capita. Kenya’s share of FDI inward stock was 55% among the East African countries in the mid 1990s but this declined to 18% by the end of 2003. The biggest beneficiary of this loss was Tanzania who’s share rose by 34% share, rising to 46% by the year 2003.

The same scenario was repeated in the period 2003-2009 where the average FDI flows into Kenya was US$ 106 million per annum compared to US$ 456 and US$ 521 million for Tanzania and Uganda respectively (World Bank, 2010). Kenya now attracts about one third of what each of her neighbours attracts in terms of FDI inflows. This situation has persisted despite the Kenya government’s attempts to implement a series of measures aimed at attracting foreign investors into Kenya since 1988, especially with respect to export platforms such as Export Processing Zones (EPZs). Nevertheless, these export platforms have themselves been disappointing in performance; with exports from EPZs accounting for about 3.5% of total manufacturing exports while employment in these firms accounted for barely 1% of total manufacturing employment by 1997 (see Glenday and Ndii, 1997). This rose somewhat due to the effects of the African Growth and Opportunity Act (AGOA) after 2001.Kenya also missed out in the global surge in FDI that was experienced in most parts of the world in the 1990s and beyond. While the average FDI inflows to Kenya doubled in the 1981-85 and 1996-2003 periods, the average inflow into African countries increased sixfold and the average inflows into developing countries as a whole increased almost tenfold. It seems clear that Kenya’s poor performance in attracting FDI at a time of global surge of inflows and with similar economic structures must be found mainly within the country.Studies on Kenya’s inability to attract FDI despite it having been the prime destination of FDI in the 1970s and 1980s have identified such factors as macroeconomic instability, corruption and



bad governance, inconsistencies in economic policies, deteriorating public service and poor infrastructure as some of the factors responsible for the low FDI inflows. These studies also highlight market size, low economic growth, lack of policy transparency and rising cost of electricity and labour. The studies include Kinaro, 2006; Opolot et al, 2008 and UNCTAD, 2005 among others. The deterioration of Kenya’s infrastructure, particularly at a time of major improvement in infrastructure in other parts of the developing world have induced many foreign investors already established in the manufacturing sector to divest or consolidate their operations out of Kenya in recent years.The trend of FDI in Kenya has shown that foreign investors are moving out of Kenya with few new investors coming in or even existing investors planning significant expansion. Kenya’s Vision 2030 asserts that the country intends to attract at least 10 large strategic investors in key agro-processing industries and raise its market share in the regional market from 7% to 15% by the year 2012. Exports can affect the economy as a whole through productivity enhancing externalities such as technology spillovers and therefore if FDI is found to promote exports, FDI can enhance economic growth. Numerous studies have concluded that exporting is crucial to growth and foreign direct investments can play a role in enhancing the export capability of a country (see Bernard et al, 2000; Bernard and Jensen, 2001; Bigsten et al, 1999, 2002; Girma et al, 2005, and Kneller et al, 2004).The Investment Promotion Act enacted in 2004 is a key policy initiative aimed at promoting foreign direct investment in the country. It provides incentives and

promotes foreign direct investments that earn foreign exchange, provide employment and promote backward and forward linkages and transfer technology. The Act, however, took away some of the benefits through imposing compulsory investment certificates and high minimum capital requirements, thus creating a legal barrier to and administrative burden for FDI thereby discouraging both domestic and foreign investment (UNCTAD, 2005). These general restrictions of the Act are contrary to practice in many other countries in Africa and elsewhere in the world that adopt more liberal entry regimes and / or more precisely targeted policies to regulate FDI entry. Tanzania for example does not impose minimum capital requirements for FDI entry in general, but makes special incentives conditional upon holding an investment license and investing a minimum of US$ 300,000 (compared to Kenya’s US$ 500,000). Uganda does not require foreign investors to invest minimum amounts but offers the facilitation support of its Investment Authority when investments exceed US$ 100,000. Additionally, the minimum capital requirement does not effectively grant protection to national investors in sensitive areas and maximize the benefits of FDI. The size of investment is by no means an indicator of “seriousness” and benefits to the economy since at times large foreign investments may crowd out small national investors’ as much as more modest foreign investments (UNCTAD, 2005). The UNCTAD(2011) data shows that Tanzania and Uganda have made tremendous improvements in their attractiveness for FDI since 1994. Tanzania’s FDI increase has been attributed to the mining sector especially uranium and tanzanite, gas and oil



discoveries as well as favourable policies that liberalized both local and foreign investments (Kajara 2010). Uganda’s FDI increase has been attributed to a wide range of tax incentives to businesses as well as its own discovery of oil and gas reserves (Ngowi, 2005). This suggests that FDI increases with increases in discovery of natural resources. Furthermore, the

UNCTAD (2011) figures suggest that economic growth rate of Tanzania and Uganda has exceeded that of Kenya since 1994 when their FDI begun to increase. This may lend credence to the hypothesis that increases in FDI leads to increases in economic growth, but these are not proportional suggesting that other factors also affect growth.

Table 5: Flows into Kenya and Tanzania (in selected years, in US$ million)

Item 2005-2007 (annual average) 2009 2010 2011 2012Kenya 267 115 178 335 259Kenya’s FDI as % of Gross Fixed Capital Formation (GFCF)

6.1 - 2.7 4.9 2.9

Tanzania 640 953 1813 1229 1706Tanzania’s FDI as % of GFCF

15.3 - 24.6 15.6 17

Source: UNCTAD (2013).



Table 6: FDI stock in Kenya and Tanzania (selected years, in US$ million)

Item 1995 2009 2010 2011 2012Kenya 732 2104 2282 2617 2876Kenya’s FDI stock as % of GDP 6.3 - 7.1 7.7 7.0Tanzania 620 8066 8762 9278 10984Tanzania’s FDI stock as % of GDP 10.2 - 37.1 38.1 38.2Source: UNCTAD (2013).

UNCTAD (2005) had argued that Kenya’s inability to attract FDI is a result of corruption, poor governance, inconsistencies in economic policies and structural reforms, deteriorating public service and poor infrastructure all of which are being addressed. Despite massive efforts by the government to implement reforms such as trade reforms, the country continues to loose its competitiveness for FDI to Uganda and Tanzania. However, Sims (2013) indicates that inflows of FDI to Kenya could match those of Tanzania and Uganda beginning 2014 aided by opportunities created by the discovery of oil deposits in Turkana. The FDI inflows are projected to average US $ 1.3 billion annually for the period 2013-2018 , placing it at par with Tanzania and Uganda who had over the years attracted more investors due to their vast natural resources such as gas , oil and other minerals.The UNCTAD (2011) figures show that Uganda and Tanzania have overtaken Kenya in terms of growth rates due to their rising FDI inflows , but Kenya is still the regional business leader ,is it that FDI is a key factor in driving economic growth or are other factors equally important ?.Some of these shortcomings have been recognized by the government and it has sought to amend the Investment Promotion Act by making investment certificates optional for all investors .The special incentives remain conditional upon

holding a certificate, though the minimum capital requirement to qualify for one would be lowered to US$ 100,000 for foreign investors and US $ 13,000 for national investors.From the year 2000, the Kenya government has implemented a number of initiatives to improve both economic performance and stimulate foreign direct investments. The government joined the Free Trade Area of the Common Market for Eastern and Southern Africa (COMESA) in 2000; negotiated for the resumption of donor aid by the International Monetary Fund (IMF); adopted the United Nations Millennium Development Goals (MDGs) in 2002 and resolved to reduce poverty levels by half by the year 2015; implemented the Economic Recovery Strategy for wealth and employment creation (ERS) in 2003 to stimulate private investment to generate wealth and reduce poverty; implemented Kenya’s Vision 2030 in 2008 and promulgated a new constitutional dispensation in 2010. The Vision, implemented in successive five-year medium term plans, with the first plan covering the period 2008-2012, is expected to encourage FDI, achieve high average Gross Domestic Product growth rate (of 10% per annum beginning 2012) and boost investments. It is also expected to enhance macroeconomic stability, raise national savings (from 17% in 2006 to 30% by 2012). These measures did see



Kenya’s growth rise from 0.6% in 2000 to 7% in 2007. The growth rate however fell to 1.6% in 2008 but has been rising since to 4.7% in 2012. The fall in GDP growth was due to global financial crisis; fall in commodity prices and post-election chaos that followed the December 2007 general elections in Kenya.Despite the recent impressive economic growth, FDI flows to Kenya average below US$ 39 per capita between 2003 and 2006 compared to US$ 418 and US$ 310 for Tanzania and Uganda respectively. By 2009, Kenya’s net FDI flows stood at US$ 116 million while Tanzania’s and Uganda’s US$ 415 and US$ 789 respectively (World Bank, 2012). This is despite the Kenyan governments implementing a series of measures to attract foreign investors that included among others Manufacturing Under Bond (MUB) in 1987, Export Processing Zones (1990) and accession to the African Growth and Opportunity Act (AGOA) in 2001 (World Bank, 2012). The last measure however led to significant FDI inflows from Asia whose investors used Kenya as a platform for quota-hopping to access the otherwise restricted US market, particularly for clothing manufactures (UNCTAD, 2005).

The Study ObjectivesIt has been argued in numerous studies that FDIs contribute positively to economic growth in the host economies. This is particularly true where FDIs bring in investible financial resources and fill the gap between desired investment and domestically mobilized savings, facilitate entry into export markets, and strengthen the export capabilities of the host country resulting in productivity gains, technology transfer, introduction of new processes,

managerial skill s and knowhow in the domestic markets, employee training, international production networks and access to markets (Caves 1998; Ayanwale, 2007; Borensztein et al, 1998. Findlay (1978) also makes a case for the increase in the rate of technical progress in the host country through a “contagion effect” from the more advanced technology as a result of FDIs. FDIs have also been credited with increase in tax revenues and improvement in management and labour skills in host countries (Todaro and Smith, 2003; Hayami, 2001). Employment creations, human capital development, contribution to international investments are some of the positive effects of FDIs (Jenkins and Thomas, 2002; World Bank, 2002).To the contrary, despite the important role played by FDI in economic growth in host countries, the level of FDI in Kenya has been low and stagnant over the past couple of decades as alluded to above. It is equally clear that FDI flows and GDP growth rates fell in the 1980s and 1990s. After 2000, rising economic growth rates contrasted with low and stagnant FDI flows. Kenya’s experience also contrasts with both Uganda and Tanzania where both FDI flows and economic growth have been on a steady rise since the early 1990s.There are few studies that analyze the empirical relationship between FDI and economic growth in Kenya; these include Kinaro, 2006; Opolot et al, 2008, and Mwega and Ngugi, 2007. Though it can be important in informing government investment policy in the host country, the empirical linkage between FDI and economic growth in Kenya is not clear. On the basis of the foregoing arguments the study will raise two main questions; namely what is the empirical relationship between Foreign Direct Investment and



economic growth in Kenya? And what factors determine the FDI flows to Kenya? The objective of the study is to empirically investigate the relationship between foreign direct investment and economic growth in Kenya and to examine and quantify the factors that drive foreign direct investment flows into Kenya. Specifically the study will seek to use Kenyan FDI flows and gross domestic product data to establish the empirical relationship between foreign direct investment and economic growth in Kenya with a view to quantify the relationship. The study also seeks to determine and empirically quantify the factors that drive FDI flows into Kenya and to suggest policy options that can be implemented to increase both FDI inflows into Kenya and hence increase economic growth in the economy based on the results of the study.The rest of the paper is organized as follows. Following this introduction, the next section briefly reviews the literature on the FDI and economic growth and their relationship as well as providing the theoretical foundation of the study; this is followed by a brief presentation of the methodology and a theoretical framework to be used in the study and includes a model to be estimated. The section also briefly discusses the types and sources of the data used in the study. The last section discusses the results and policy implications based on the results of the study.Literature ReviewIn this section we briefly review some of the theoretical and empirical literature on foreign direct investments and economic growth. The section is divided into three parts comprising theoretical literature review, empirical review and an overview of the literature.

Theoretical FoundationIt has been argued that foreign direct investment can either positively or negatively affect economic growth in the host economies. There are many channels through which FDI can impact on growth. Blomstrom et al (1994) argue that FDI exerts a positive effect on growth but there is a threshold level of income above which FDI has a positive effect on economic growth and below which it does not or is insignificant. Borensztein et al (1998) is of the opinion that the interaction of FDI and the quality of human capital has important effect on economic growth and suggests that the differences in the technological absorptive ability may explain the variation in growth effects of FDI across countries .They point out that countries may need a minimum threshold of human capital to experience positive effects of FDI on economic growth. It is similarly suggested by Oloffsdotter (1998) that the beneficial effects of FDI are stronger in the countries which have a higher level of institutional capability.

In another study Alfaro et al (2003) find little support that FDI has an exogenous positive effect on economic growth, however, their findings suggest that local conditions such as the level of education and the development of local financial markets play an important role in allowing the positive effects of FDI to materialize in the economy.Many studies exist concerning FDI and its determinants, where the main factors include the rate of investment return, market size, macroeconomic variables (especially their stability), quality of labour, infrastructure, property rights and the effects of globalisation which has led



to FDIs in LDCs to shift from market seeking and resource seeking to the more efficiency seeking FDIs.The literature indicates that there is no conclusive argument on factors determining FDI and its subsequent effect on economic growth. Causality between FDI and growth is still unclear. The direction of FDI might be associated with domestic policy variables. The direction of the relevant causalities between FDI and growth may well depend on the determinants of FDI. If the determinants have strong links with growth in the host country, growth may be found to cause FDI, and output may grow faster when FDI takes place in other circumstances like the case of oil discovery.The understanding of the impact of specific categories of foreign capital inflows has important policy implications. Most studies on the growth of specific types of foreign capital flows focus on FDI. However, the empirical evidence on FDI and its impact on host countries growth are ambiguous at both the micro and macro level. The positive effects on the growth of the host economy can come from investible financial resources filling the gap between investment and domestically mobilized savings, facilitation of entry into export markets and strengthening export capabilities of the recipient country. Caves (1998) has postulated that other positive effects of FDI include productivity gains, technology transfer, new processes, managerial skills, employee training, international production networks and access to new markets. Borensztein et al (1998) see FDI as an important channel for transfer of technology and contributing to growth in larger measure than domestic investment. Findlay (1978) postulates that

FDI increases the rate of technical progress in the host country through what he calls a “contagion effect” from the more advanced technology, superior management practices used by the foreign firms. Todaro and Smith (2003) and Hayami (2001) noted that FDI may also increase tax revenue, improve management, technology as well as labour skills in host countries. Many other studies have noted the benefits of FDI to include new technology, employment creation, human capital development, international trade integration, enhancing domestic investment, and increased revenue (Jenkins and Thomas, 2002; World Bank, 2000). FDI is seen as being a positive contributor to the economic growth of the host country.However, it has also been argued that foreign direct investments can also have adverse effects on the economy of the host country. Reis (2001) has advanced the argument that opening up a country to FDI in the research and development sector may replace the domestic firms and decrease welfare due to the transfer of capital returns to foreign firms.Firebaugh (1992) points out that the foreign firms may fail to encourage local entrepreneurship, reinvest profits, develop linkages with domestic firms or fail to use appropriate technology. FDI can be detrimental if it “crowds out” domestic businesses and engenders inappropriate consumption patterns or reduce domestic savings and investment rates by stifling competition through exclusive production agreements with the host country. FDI may also lead to less than optimal corporate taxes where they are provided with liberal tax concessions and excess investment allowances and other incentives. In a distorted market, FDI can



lead to negative value-added at world prices coupled with repatriation of profits and dividends (Mwega and Ngugi, 2007). The study is based on the theory of profit maximisation, such that the country’s GDP can be increased by the input of the FDI inflows which enhances its productivity as it helps local firms to be more productive through the infusion of capital and more modern and efficient technologies that it brings as well as fostering competition both locally and in the country’s foreign markets.It is generally believed however, that FDI provides net benefits to the host country. This explains why the importance of FDI in economic performance has been extensively discussed in the economic literature.

Empirical StudiesSeveral studies have been conducted on the empirical relationship between FDI’s and economic growth. Some of these studies have shown that FDIs positively influence economic growth in the host countries. Dees (1998) in a study on the determinants and effects of foreign direct investments in China found that FDI has been important in explaining China’s economic growth. Similarly, de Mello (1997) also presents a positive correlation between FDIs and economic growth of selected Latin American countries. Barrel and Pain (1999) explored the benefits of FDI of U.S multinational in four European Union countries and find that FDI may affect the host country’s performance positively in the case where there are transfers of technology and knowledge through the FDI to the host economy.

Firm-level studies of specific countries provide contradictory evidence on the role

played by FDI in economic growth. For example Wilmore (1986) examining a sample of 282 pairs of firms from 80 industries in Brazil found that FDI had a beneficial impact on growth since foreign firms are more efficient than domestic ones. Moreover, Blomstrom (1986) found that FDI enhances productivity growth of Mexican firms. FDI spillovers that occur when the entry or presence of a foreign investment firm(s) contribute to the productivity or efficiency benefits of indigenous firms are critical in defining the impact of FDI on the growth of host nations.The literature identifies competition, linkages, labour mobility, skills and imitation as the main channels of technological spillovers from FDI to indigenous firms (Blomstrom and Kokko, 1998). We however, note that FDI spillover may either be positive or negative on their impact on economic growth in the host countries. Some empirical studies show positive effects of FDI spillovers on economic growth in the host countries.Some empirical studies show positive effects of FDI spillovers on economic growth (Caves, 1974) on Australia and Kokko (1994) for Mexico. However, Haddad and Harrison (1993) find no evidence of positive spillovers from FDI in Morocco. The study by Aitken and Harrison (1999) for Venezuelan firms in the period 1979-1989 and Djankov and Hoekman (2000) for the Czech Republic firms report negative spillovers. Hanson (2001) concludes that the evidence that FDI generates positive spillovers for host countries is weak. Microeconomic studies report positive effects of FDI and productivity spillovers, these include studies by Lipsey and Sjöholm, 2004;



Black and Gertler, 2008. Most macroeconomic studies generally suggest that FDI exerts a positive impact on economic growth in particular contexts. Balasubramannyam et al (1996) and Zhang (2001) find that effects on growth of FDI are more significant in the presence of trade openness and where host country adopted liberalisation. Borensztein et al (1998) argue that FDI is an important channel for the transfer of technology and contributes to economic growth when the country has a highly educated workforce. Blomstrom et al (1994) found that among developing countries, the positive impact of FDI on growth is larger in those countries that exhibit higher levels of per capita income. FDI is also beneficial for economic growth when the country has sufficiently developed and sophisticated financial markets (Alfaro et al, 2004). The other factors that enable FDI to positively impact on growth include political and economic stability as well as the quality of institutions and infrastructure which complements FDI (see Oloffsdotter, 1998; Hall and Jones, 1999; Rodrik et al, 2002; Aschauer, 1989 and Tondl and Prüfer, 2007). The literature therefore suggests that openness to trade, human capital, financial market development, public infrastructure and quality of institutions affects a host country’s ability to absorb FDI spillover.The literature on FDI shows that its impact on economic growth can either be direct or indirect. The indirect impact or spillovers are dependent on the host country’s conditions. Specifically this depends as per capita income, human capital stock, financial sectors level of sophistication, the level of development and quality of public infrastructure, the quality of institutions, trade openness and

macroeconomic stability. The empirical evidence however shows that the relationship between FDI and growth is uncertain and varies across host countries. This paper proposes to use Kenyan data to find out whether FDI enhances economic growth in Kenya. Using panel data for 25 central and Eastern European and former Soviet transition economies, Campos and Kinoshita (2003) examined the effects of FDI on growth for the period 1990-1998. Their main results indicated that FDI has a significant positive effect on economic growth of each country. Focusing on the factors that explain growth in developing countries, Blomström et al (1994) found that foreign direct investments exerts a positive effect on economic growth but that there seems to be a threshold level of income above which FDI has positive effect on economic growth and below which it does not. The explanation is that only those countries that have reached a certain income level can absorb new technologies and benefit from technology diffusion, and thus reap the extra advantages that FDI offer. They concur with other studies that suggest human capital as one of the reasons for the differential response to FDI at different levels of income (Borensztein et al, 1998; Bengos and Sanchez-Robles, 2003).However, some studies have found that FDI may not influence long-run economic growth. In a study on the interaction between foreign direct investment, economic freedom and growth, Bengos and Sanchez-Robles (2003) estimated the relationship between FDI and economic growth using panel data for eighteen Latin American countries over the period 1970-1999. They show that FDI had positive and significant impact on economic growth. However, they also found that the



host country requires adequate human capital, political and economic stability and liberalized market environment so as to gain from long-term FDI inflows. It has also been shown by Ang (2008) that better developed financial systems allow an economy to exploit the benefits of foreign direct investment more efficiently. The author used Thailand as a case study to examine the role of FDI and financial development in the process of economic development. The estimation uses an unrestricted error-correction model to avoid omitted lagged variable bias, and an instrumental variable to correct for endogeneity bias. Using annual firm series data from 1970-2004 the results show that financial development stimulates economic development whereas foreign direct investment impacts negatively on output expansion in the long-run. However, an increased level of financial development enables Thailand to gain more from FDI, suggesting that the impact of FDI on output growth can be enhanced through financial development.Some studies indicate that the relationship between the FDI and economic growth is weak and insignificant. Ayanwale (2007) investigating the empirical relationship between non-extractive FDI and economic growth in Nigeria using annual time series and ordinary least squares technique found the relationship between FDI and economic growth to be positive but not significant.

Overview of The LiteratureIt is clear from this brief review that the effects of FDI on growth are dependent on the characteristics of the host country and the sectors where the FDI is directed. Large market size and high incomes may attract market seeking FDI as opposed to

small and low income economies. Market-seeking FDI is therefore induced by market access to host countries for efficient utilization of resources and exploitation of economies of scale. FDI complements growth when directed towards highly productive sectors of the host economy. Investment in development of good quality infrastructure, low cost and highly skilled human capital and innovations and technological progress increase productivity and promote growth in the long-run. The empirical evidence shows that the relationship between FDI and growth and the expected significance of the determinants of FDI varies across host countries.

The empirical evidence shows that the relationship between FDI and growth and the expected significance of the determinants of FDI varies across host countries. Some studies show positive effect of FDI on economic growth while others show negative impact; whereas some studies exhibit insignificant relationship between the two variables. The differences in empirical results may be attributable to differing economic institutional and technological conditions in the recipient countries. The few country specific studies also indicate that there may be endogenous relationship between FDI and growth which may have to be taken into account if the results are to be robust. The next section presents the methodology used in the study.

Theory And Empirical ModelIn this section we present the methodology used in the study. The methods are outlined before the model is specified.



Section 3.2 briefly presents the types, sources and measurement of the data used. The relationship between FDI and economic growth is situated in growth theory that pronounces the role of improved technology, efficiency and productivity in promoting growth (Lim, 2001). However, the potential contribution of FDI to growth is dependent strictly on the on the circumstances in the recipient or host countries. Certain host country conditions are necessary to facilitate the spill-over effects.In this study, we assume in line with standard economic theory that foreign capital inflows into a recipient country will increase its stock of capital and level of technology and lead to better economic performance. Foreign direct investment will affect economic growth positively through improved technology, efficiency and increased productivity (Lim, 2001). However, as noted in the review of literature, the potential contribution of FDI to growth is strictly dependent on the circumstances in the recipient or host country. Theoretically a country’s GDP is influenced by a variety of factors. RGDP is measured as nominal GDP deflated by the GDP deflator, (base 2000 s=100).We hypothesise FDI to positively influence growth. FDI promotes economic growth in host countries by providing external capital (O’hearn, 1990). Apart from increasing tax revenues, improving managerial and labour skills it also creates employment in the host country. The gross fixed domestic investment (GFDI) is proxied in this study by the share of the gross domestic capital formation to GDP less net FDI flows. Increased investments rates promote productivity in a country as argued by Grossman and Helpman (1991). We expect a positive relationship between

this variable and GDP. The level of human capital measured here by the secondary school enrolment rate should have a positive impact on GDP. It can be argued that widespread availability of cheap and highly skilled labour force tends to attract private investment and enhances productivity in a country (Akinlo 2004, Barro and Lee, 1994). The rate of inflation measured here by the annual percentage change in consumer price index is a reflection of macroeconomic stability. A low and stable inflation rate implies a more reliable economic environment enabling investors to benefit from existing opportunities (see Larrain and Vergara, 1993; Serven and Solimano, 1993). A negative relationship exists between inflation and GDP. The size of Government (GOVSIZE) is measured in this study by the share of total government consumption to GDP. Anyanwale (2007) points out that higher level of general final government consumption provide social capital that encourages production and growth. Trade openness (OPEN) on the other hand promotes economic growth through increasing competitiveness and providing access to markets for finished products, (Balassa, 1978). It also enables the importation of raw materials and capital goods and facilitates access to new technologies and skills. Large export markets encourage inward market-seeking FDI and foreign capital inflows (Kinaro, 2006; Ajayi 2007).There are evidently other influences on GDP with varying degrees of importance are captured by the error term (e) in our specification.



The Model As explained in the section above, equation 1 below shows the hypothesized relationship between economic growth and its determinants.We hypothesize economic growth to be influenced by several factors as shown in the model belowRGDP= f ( FDI ,GFDI ,UMCAP , INFL, GOVSIZE ,OPEN , ROI , INFRAC ,TDS,CPI ). .. .. . .. .. . .. .. .(1)Where

RGDP = Real Gross Domestic ProductFDI = Foreign Direct InvestmentGFDI = Gross Fixed Domestic InvestmentHUMCAP = Level of Human CapitalINFL = Rate of InflationGOVSIZE = Government consumptionThe equation which relates the real gross domestic product to various factors that influence it can be elaborated as

RGDP=g0+g1 FDI+g2 HUMCAP+g3 INFL+g4GOVSIZE+g5 OPEN+g6+ε ). .. .. . .. .. . .. .. .(2)

Where g1,....., g6 are the coefficients to be estimated and g0 is the constant. ε is the error term.Equation (2) above shows the growth model to be estimated.When estimating the growth equation it is possible that some of the variables could be correlated to the error term resulting in a problem of endogeneity and could give

rise to biased estimated coefficients. If this were to happen appropriate instruments will be searched and used in a 2 stage least squares (2SLS) estimation.Since we also wish to quantify the factors that drive direct foreign investment (DFI) inflow into the country we shall estimate an FDI equation. We hypothesize that direct foreign investment is influenced by a variety of factors as shown below:

FDI=ℜ0+ℜ1 RGDP+ℜ2 INFRAC+ℜ3OPEN+ℜ4GR+ℜ5RLIR+ℜ6TDS+ℜ7 ROI +μ. . ..(3 )

WhereFDI = Foreign Direct Investment measured share of FDI to GDPRGDP = Real Gross Domestic Product (nominal GDP deflated by the GDP deflator

HUMCAP = Level of Human Capital (proxied by the secondary school enrolment rate)GR = Market size (measured by annual % change in real GDP)RLIR = Real Interest rate (measured by the difference between the nominal

lending interest rate and the rate of inflationGFDI = Gross Fixed Domestic Investment (proxied by the share of the gross domestic capital formation to GDP less net FDI inflows)OPEN = Openness of the economy (measured by the ratio of trade exports + imports to GDP)INFRAC = Infrastructure (proxied by the electric power transmission and distribution losses as a % of the total output)INFL = Inflation rate measured by the annual % change in consumer price index



TDS = Total debt service to GDP ratio measured by the share of total external debt service to GDPCPI = Corruption Perception IndexROI = Return on Investment proxied by long-term US interest rates GOVSIZE = Government consumption measured by the share of the total government consumption to GDPµ = The Stochastic error term

The coefficients ℜ0 .. .. . .. .. . .. .ℜ7 are to be estimated.In the FDI model equation (3) above, FDI is measured as the ratio of FDI to GDP and is the dependent variable. It is hypothesised that a high real GDP reflects large market size that attracts further FDI especially the market seeking ones, resulting in more demand for products or services to be provided by FDI (Chunlai, 1997; Mwega and Ngugi, 2007). The Gross Fixed Domestic investment increases the rate and efficiency of domestic capital investment, raising productivity in a country and thereby encourages FDIs. We expect a positive impact on FDI. The measure of openness is as defined above. This may encourage exports and hence lead to market seeking FDIsInfrastructure is critical for both economic growth and competitiveness. In this study it is proxied byElectric power transmission and distribution losses as a percentage of total output. We expect this to have a negative impact on FDI inflows as it relates to high cost of production (see Anyanwale, 2007).The real exchange rate has an important impact on FDI inflows. A depreciation of the exchange rateEncourages higher inflows as it makes local assets and production costs cheaper.

An appreciation of the exchange rate has the opposite effect.The choice of variables included in the model specifications has been guided by the theories of economic growth and the determinants of FDI inflows discussed in the literature review above.The following section elaborates the sources and types of data used in the study.

Data Sources, Types and Measurement

The study covers the period 1970-2010; and therefore includes the period which Kenya was the preferred FDI destination in East Africa as well as the period in which she was overtaken by both Tanzania and Uganda as the main FDI destinations in the region.

The data is annual time series data obtained from secondary sources. The sources include the Central Bank’s annual economic reviews, Republic of Kenya Statistical Abstracts and economic surveys produced by the Kenya National Bureau of Statistics (KNBS). Other sources of data included the World Bank’s World Development Indicators and Global Development Finance reports. The UNCTADs World Investment reports and the IMF’s International Financial Statistics have also been used. Due to difficulties in obtaining certain qualitative data on such variables like corruption it was left out of the analysis even though we are acutely aware that corruption levels in a country can have a major negative impact on the inward flow of FDIs.

Time Series Properties and Estimation TestsGiven the time series nature of our data, it was imperative to carry out estimation



tests to be sure that our data is not non-stationary so that we avoid the problem of spurious regression results. We therefore conducted stationarity tests for the series using the Augmented Dickens Fuller (ADF) test. The ADF assumes that the error terms are independently and identically distributed. A time series data is said to be stationary if its mean, variances and autocovariance remain the same no matter at what point we measure them.

Unit root test for stationarity resultsWe used the Augmented Dickens Fuller test to test for stationarity in our data. The

test indicates whether or not the variables are stationary. The null hypothesis is that of non-stationarity while the alternative hypothesis is that of stationarity. The test statistic is then compared with the t-critical. If the t-statistic is less than t-critical we reject the null hypothesis of non-stationarity and therefore the series is stationary. On the other hand, if the t-statistic is more than the critical we accept the null hypothesis and non-stationarity and the series is therefore non-stationary and prone to spurious regression. The table below shows that our data was stationary and we do not face the possibility of spurious regression results.

Table 7: Unit Root Test using ADFVARIABLE ADF

STATISTIC1% CRITICAL VALUE

5% CRITICAL VALUE

10% CRITICAL VALUE

NATURE

RGDP -6.900 -3.648 -2.958 -2.612 STATIONARYGR -5.015 -3.648 -2.958 -2.612 STATIONARYFDI -6.941 -3.648 -2.958 -2.612 STATIONARYHUMCAP -6.426 -3.648 -2.958 -2.612 STATIONARYOPEN -6.303 -3.648 -2.958 -2.612 STATIONARYINFRAC -5.139 -3.648 -2.958 -2.612 STATIONARYGOVSIZE -5.190 -3.648 -2.958 -2.612 STATIONARYTDS -6.008 -3.648 -2.958 -2.612 STATIONARYINFL -6.352 -3.648 -2.958 -2.612 STATIONARY

RLIR -5.772 -3.648 -2.958 -2.612 STATIONARYROI -6.761 -3.648 -2.958 -2.612 STATIONARY

Results in table 5 above show that the ADF test statistics are less than the t-critical at the 1%, 5% and 10% and we therefore reject the null hypothesis of non-stationarity and accept that the series are stationarity and our OLS regression could be conducted since the results would not be spurious (The OLS results are shown in tables 4 and 5).The study hopes to determine the important variables that may be important

in encouraging the inflow of direct foreign investment to Kenya and recommend policies that can enhance the inflow of FDI into Kenya. The study also hopes to determine the empirical relationship between economic growth and foreign direct investment in Kenya with the view to both boost inflow of direct foreign investment and economic growth.Having elaborated the hypothesised relationships between our dependent



variables and the independent variables in the two models we estimated the relationships using time series data and the results are discussed in the following section.

Results and DiscussionWe set out to empirically investigate the relationship between foreign direct investment and economic growth in Kenya by examining the factors that drive foreign investment flows into Kenya. The objective was to establish the empirical

relationship between FDI and economic growth in Kenya.

Regression ResultsThe growth equation

From our results in table 1 we see that the growth in GDP is positively influenced by human capital and the variable is significant at the 1% level (t-value=4.96) and a P- value=0.000. The results also show that the government expenditure (GOVSIZE) is a significant determinant of the real GDP.T he variable is significant at the 1% level with a t-value of 3.17 and a p-value of 0.003.

Table 8: Determinants of RGDP

Variables Coefficient Standard Error t-values P-valuesfdi 1949.496 2514.297 0.78 0.443humcap 822.911 166.0322 4.96 0.000infla 9.835928 32.06965 0.31 0.761govsize 3523.373 1111.173 3.17 0.003open 585.8971 215.6161 2.72 0.001_cons -45026.6 23981.22 -1.88 0.069

No. of obs = 41; F(6, 34)= 21.28; Prob>F = 0.0000; R-Squared = 0.7897; Adj R-Squared = 0.7526; Root MSE = 8237.2

It was argued that the government size which is measured by the share of total government consumption in GDP should influence economic growth in a positive manner. The higher the level of the general final government consumption the more the social capital and this encourages production and the growth of GDP. The results indicate a highly significant coefficient (t-value=3.17) at the one percent level of significance. The results therefore suggest that government consumption is a major contributor to GDP growth and should be encouraged. It appears that government expenditure on social services and other amenities is an important boost to growth.

The openness of the economy to the rest of the world is similarly shown to be a major driver of the GDP growth rate. The variable was measured as the ratio of trade defined as imports plus exports to the GDP. The idea here is that trade openness promotes economic growth through increasing competitiveness and providing access to markets both for finished products and exports as pointed out by Balassa (1978). By enabling importation of raw materials and capital goods and facilitating access to new technologies and managerial skills it positively impacts on the growth of GDP. The coefficient is large and positive and is reported to be



significant at the 1% level (t-value=2.72) with a p-value of 0.010.The variables in the growth equation explain about 80% of the variations in the GDP growth rate and the overall model seems to be well specified with an F-statistic of 21.2. The p-value for the model as a whole indicates it is fairly well specified.

The FDI equationThe results in our table 2, shows the determinants of FDI inflows to Kenya. The real gross domestic product measured as defined above is shown to be a major influence on the FDI inward flow.



Table 9: The determinants of FDI

Variables Coefficient Standard Error t-values P-valuesrgdp 2.69E-05 1.09E-05 2.47 0.019infrac 0.066262 0.03405 1.95 0.06open 0.017142 0.013785 1.24 0.223gr -0.00854 0.022764 -0.37 0.71rlir -0.00668 0.010635 -0.63 0.534tds 0.072952 0.043933 1.66 0.107roi -0.07517 0.05979 -1.26 0.218_cons -0.73934 1.102412 -0.67 0.507

No. of obs = 41; F(8, 32)= 1.86; Prob>F = 0.1011; R-Squared = 0.3179; Adj R-Squared = 0.1474; Root MSE = .52994

The theoretical basis for this is that a high real GDP reflects a large market size that attracts FDI, especially the market seeking type. The high GDP leads to higher demand for the products and services provided by the foreign firms. A high real GDP therefore has a positive influence on the FDI. The results in table 2 show that the variable is highly significant at the 1% level and has a t-value of 2.47. The coefficient is positive as was hypothesized. The results further show that the infrastructure is also an important influence on the FDI inflow. The variable has a positive coefficient and is significant at the 10% level with a t-value of 1.95. We had proxied infrastructure by electric power transmission and distribution losses as a percentage of total output. The expectation was that it would have a negative impact on FDI inflows due to the measurement method employed as it relates to high cost of production. However, the results do not support this hypothesized relationship. It is possible and even likely that good infrastructure which improves a country’s competitiveness may in fact attract FDI

flows due to improved cost conditions in a country.The variable on total external debt service as a ratio of GDP (tds) captures the liquidity and solvency constraints imposed by the debt burden and the higher the debt services ratio the more it deters FDI inflows. The results indicate that the coefficient is positive though not statistically significant. The a priori expectation was that it would be negative due to the negative impact of the debt burden which discourages FDI flows.These results imply that Kenya can attract more FDI which are acknowledged to have potential benefits that can accrue to the country. FDI is important for economic growth as it provides the much needed capital for investment, increases competition within the country and aids local firms to become more productive by adopting more efficient technology or by investing in human or physical capital. FDI also contributes to growth in a substantive manner because it is more stable than other forms of capital flows.



Conclusion And Policy ImplicationsThe driving objective of this study was to investigate the empirical relationship between Foreign Direct Investment and economic growth in Kenya. The study also set out to empirically investigate the factors that drive FDI flows into Kenya, having established that Kenya’s FDI inflow record over the recent years has not been impressive despite her being among the most favoured FDI destinations in the 1970s in Eastern Africa. The realisation that Kenya is now among the countries with very low levels of FDI motivated the study. By establishing empirically which factors drive both growth and FDI inflows to Kenya, it is possible to design policies that can attract the flows into Kenya.The study has shown that human capital, government expenditure and openness of the economy are vital for the growth of the economy and therefore policies that can

enhance these factors would be needed. Furthermore, the drivers of Foreign Direct Investment have been shown to be the real GDP growth, low levels of indebtedness and improved infrastructural facilities. It is clear that the role of government would be crucial in encouraging FDI inflow to Kenya. It can also be argued that most FDI to Kenya is the market-seeking type and this requires a rapidly growing real GDP. The formation of regional blocks, and political stability would also be crucial. It can be concluded that the main FDI determinants in Kenya are market size (rgdp), political stability (d2007) and openness of the economy as well as infrastructure. The major impediments to FDI inflow would be political instability, institutional factors as well as crime and insecurity. There is little doubt that the cost of doing business, bureaucratic red-tape, high cost of electricity and other utilities and poor investment code as well as endemic corruption are some of the factors that hep explain the low FDI inflows into Kenya in the recent decades.



ReferencesAng J.B (2008), “Foreign Direct Investment

and its Impact on the Thai Economy: The Role of Financial Development”, Monash University, mimeo

Ayanwale A.B (2007), “FDI and Economic Growth: Evidence from Nigeria”, AERC Research Paper, African Economic Research Consortium, Nairobi.

Barrell, R., and N. Pain (1999), “Domestic Institutions, Agglomeration and Foreign Direct Investment in Europe”, European Economic Review, 43:925-934

Bengos, M., and Sanchez-Robles (2003),”FDI, Economic Freedom and Growth: New Evidence from Latin America”, European Journal of Political Economics, 19: 529-545

Bernard, A.B., and Jensen, J.B (2000), “Exporting and Productivity”, mimeo, Yale University

Bernard, A.B., and Jensen, J.B (2001), “Why Do Some Firms Export?”, NBER Working Paper (WP 8369)

Bigsten, A., P. Collier, S. Dercon et al (1999), “Exports of African Manufacturers: Micro policy and Firm Behaviour”, Journal of International Trade ad Economic Development, 8: 53-71.

Bigsten, A., P. Collier, S. Dercon et al (1999), “Investment in Africa’s Manufacturing Sector”, Oxford Bulletin of Economics and Statistics, Vol. 16 No. 4 pp 489-512.

Blomström, M., R. Lipsey and M. Zegan (1996), “Is Foreign Investment the Key to Economic Growth?”, The Quarterly Journal of Economics 111:

269-276

Blomström, M., and H. Persson (1983), “Foreign Investment and Spillover Efficiency in an Underdeveloped Economy: Evidence from Mexican Manufacturing Industry”, World Development, Vol. 11 pp 493-501

Borensztein, E., de Georgio, J.W. Lee (1998), “How does Foreign Direct Investment affect economic Growth?” Journal of International Economics 45: 115-135.

Campos, N.F., and Y. Kinoshita (2003), “Why Does FDI Go Where It Goes? New Evidence from the Transition Economies”, IMF Working Paper, Washington D.C., IMF Institute

Caves, R.E. (1978), “International Corporations: The Industrial Economics of Foreign Investment”, Economica 38: 1-27

Dees, S. (1998), “Foreign Direct Investment in China: Determinants and Effects”, Economics of Planning, 31:175-194

de Mello, Luis R. (1997),” Foreign Direct Investment in Developing Countries and Growth: A Selective Survey”, The Journal of Development Studies, 34: 1-34.

Findlay, R. (1978), “Relative Backwardness, Direct Foreign Investment, and the Transfer of Technology: A Simple Dynamic Model”, The Quarterly Journal of Economics 92(1): 1-16

Firebough, G. (1992), “Growth Effects of Foreign and Domestic Investment”, American Journal of Sociology 98(1): 105-130.

Girma, S., R. Kneller and M. Pisu (2005),



“Acquisition FDI and the export intensity of multinational Firms”, University of Nottingham Globalization Productivity and Technology Research Paper, RP 2005/11.

Glenday and Ndii (1997), “Export Platforms in Kenya”, mimeo

Görg, H., and Strobl (2001), “Multinational Companies and Productivity Spillovers: A Meta analysis”, Economic Journal Vol. 111, pp F723-F739.

Hanson J. (1996), “Human Capital and Direct Investment in Poor Countries”, Explorations in Economic History, 33:86-106

Hayami, Y. (2001), “Development Economics: From the Poverty to the Wealth of Nations”, Oxford University Press.

Jenkins, Carolyn and Lynne Thomas (2002), “Foreign Direct Investment in Southern Africa: Determinants, Characteristics ad Implications for Economic Growth and Poverty Alleviation”, Final Report, October, Globalization and Poverty Project, Centre for the Study of African Economies, Oxford University.

Kinaro, O.E. (2006), “Determinants of Foreign Direct Investment in Kenya”, Thesis, African Institute of Economic Development and Planning (IDEP), United Nations.

Kinuthia, B.K. (2010), Determinants of Foreign Direct Investment in Kenya, paper submitted for the Annual International Business and Management (AIBUMA Conference, Nairobi)

Kneller, R., and M. Pisu (2004), “Export-oriented FDI in the UK”, Oxford Review of Economic Policy, Vol. 20, no. 3, pp 424-39.

Lim, E. (2001), “Determinants of and Relationship between Foreign Direct Investment and Growth: A Summary of Recent Literature”, IMF Working Paper No. 175, International Monetary Fund, Washington D.C.

Mwega, F.M., and R.W. Ngugi (2007), “Foreign Direct Investment in Kenya” in Ajayi, S.I. (eds) Foreign Direct Investment in Sub-Saharan Africa: Origins, Targets, Impact and Potential, African Economic Research Consortium, Nairobi.

Opolot, J., Mutenyo, J., and A. Kalio (2008), “Determinants of Foreign Direct Investment: Evidence from Sub-Saharan Africa using A Generalized Method of Moments Dynamic Panel Estimator”, Research Bank of Uganda.

Reis, A. (2001), “On the Welfare Effects of Foreign Investment “, Journal of International Economics 54: 411-427.

Republic of Kenya (2007), “Vision 2030, A Competitive and Prosperous Kenya”, Government Printer, Nairobi

Todaro, M.P. and S.C. Smith (2003), “Economic Development”, Pearson Education Limited

UNCTAD (2011) “World Investment Report”, Ne York, United Nations Publication

UNCTAD (2013), World Investment Report: Global Value Chains, Investment and Trade for Development

UNCTADstat. Available at: http://uncstadstat.unctad.org


http://uncstadstat.unctad.org/


United Nations Conference on Trade and Development (1998), “World Investment Report“, New York, United Nations Publication

United Nations Conference on Trade and Development (2005), “An Investment Guide to Kenya”, Opportunities and Conditions, New York, United Nations Publication.

World Bank (1996), “World Debt Tables: External Finance for Developing Countries”, Vol. 1 (Analysis and Summary Tables) Washington D.C.,

The World Bank

World Bank (2000), “World Business Environment Survey”, Washington D.C., World Bank.

World bank (2012). Connecting to Compete: Trade Logistics in the Global Economy, Washington D.C.

World Investment Report (2006), “FDI from Developing and Transition Economies Implications for Development”, U.N.


chapter onechss.uonbi.ac.ke/sites/default/files/chss/vol 4 no 1.docx · web viewalthough the...

Documents