Effect of the Migration Option on Rural Microenterprise Establishment Decisions in the Mfantsiman Municipality of Ghana

Introduction

The option to move from rural communities to other communities or cities perceived to be relatively endowed with better job opportunities may affect the decision of rural dwellers to establish a farm or non-farm enterprise in their communities. As shown by De Brauw et al. (2002), choices have shifted over time towards increased migration as a prevalent option among off-farm activities. Litchfield and Waddington (2003) indicated that migration, motivated by job search, is a common feature of household coping behaviour in Ghana. A decision-maker may thus consider local enterprise feasibility and returns against perceived returns upon migration. Ewusi (1986) found depressing social conditions prevailing at rural communities as more compelling reasons for rural people to migrate than economic factors. These findings may explain the increasing number of porters and street vendors of both sexes in the major cities in Ghana, some of whom are migrants from rural communities with limited socio-economic opportunities. Many of them are youth who complete Junior High School with their minds set on moving abroad as a first option, based on their perceptions of better social amenities and job opportunities. Twumasi-Ankrah (1995) stresses that the motives and characteristics of rural–urban migrants should be studied. Some rural migrants may not have resources or skills, nor motivation, to stay in their communities to engage in any farm or off-farm economic venture.

Several researchers have espoused the determinants and impacts of rural out-migration. Wirawan (2008) identified education, age, marital status, family’s socio-economic condition, the success of former migrants and family support from their homeland as migration determinants. Stark (1980), Hoddinott (1994) and Chen (2014) also identified determinants of migration, which include education, age, current earnings and household characteristics as well as the poverty status of the community. Todaro (1969) showed that potential risk-neutral migrants decide to migrate from rural to urban areas with an expected income maximisation objective due to wage differentials between their origin and destination. Rural–urban migration is selective of the young, versatile and better educated members of the community (Boakye-Yiadom and Mackay 2006; Ritchey 1976), which may result in a reduction in rural productivity.

Given the importance of small businesses to national economic growth, an understanding of rural dweller perceptions of the migration option, among other decision factors, is essential for microenterprise entry in rural settings. Improperly assessed factors of local development may lead to low rural enterprise establishment opportunities. Arthur (2004) revealed that authorities in the Mfantsiman Municipality in the Central Region of Ghana (separated into Mfantsiman West Municipality and Ekumfi District since June 2012) had not been effective in promoting its general development and recommended a more effective response to the pressing poverty, unemployment and rural–urban migration.

The Mfantsiman Municipality (Figure 1) stretches 21 km along Ghana’s coastline from latitude 5°7′ to 5°20′ North of the equator and longitude 0°44′ to 1°11′ West of the Greenwich Meridian and for about 13 km inland. It covers a total area of 612 km². Based on an average annual inter-censual population growth rate of 3.1 per cent in 2010 (Ghana Statistical Service [GSS] 2005, 2013), the Municipality’s population in 2013 (at the time of fieldwork) was estimated at 208,939 comprising 45.3 per cent males and 54.7 per cent females. The rural population formed 49.6 per cent of the total population (GSS 2013). The predominant spoken language in the municipality is Fante, the local dialect.

OPEN IN VIEWER

Figure 1.

Source: Department of Geography and Regional Planning, University of Cape Coast, 2013, upon request by the authors (at a fee) with the right to use it for publications.

Among the population who are 11 years or older in the municipality, 74.5 per cent were literate while 25.5 per cent were not. The higher rural non-literacy rate (28.4%) in the municipality was a primary development concern. It was much higher than the regional average of 21.8 per cent as well as the national average of 25.9 per cent. Out of the municipality’s population 15 years and older totalling 118,117, about 4.1 per cent were unemployed, which is higher than the regional average of 4.0 per cent and the national average of 3.8 per cent (GSS 2013).

This study is conducted in response to the dearth of knowledge on the interactive effect of the attributes of individuals and the opportunities provided by prevailing socio-economic factors on rural microenterprise decisions. The findings will aid policymakers and development practitioners to introduce microenterprise interventions to reduce youth unemployment and migration in the Mfantsiman Municipality. The general objective of this article is to examine the effect of migration option on microenterprise establishment decisions of rural dwellers in the Mfantsiman Municipality. The specific objectives are to:

determine how the option to migrate affects the probability of microenterprise establishment among various age groups and

Theoretical and Conceptual Framework

The broad theoretical framework of the study is decision theory. The objective of decision-making is utility maximisation for which the probabilities associated with their occurrence are evaluated for every decision-maker (Mills 1906). Descriptive decision theory provides a strong foundation for this study, which sets out to investigate how rural community dwellers make enterprise establishment decisions in the presence of a migration option. It is about how decisions are made in the real world and employs psychological analysis to explain or predict the actual action of a decision-maker (Hansson 1994). In contrast, normative decision theory describes how decisions should be rationally made and is assumed to agree with the fundamental beliefs of a rational decision-maker.

The decision-maker faces the dichotomous choice: to establish or not to establish a rural microenterprise under the influence of the migration option as the primary factor. The model for determining the importance of the migration option in enterprise establishment decisions is based on the theory of utility maximisation. Decision-makers are assumed to be seeking to maximise utility from the combination of opportunities, resources and their attributes. Under the theorem, a conjoint measurement method was used to elicit individual utility functions. Conjoint analysis is based on the principle that a product can be described by its characteristics or attributes, combinations of which may obtain different utilities. The threshold decision theory also applies to the study context. The rural decision-maker’s reaction is determined by the decision determinants, reaching a critical cut-off point or utility level for enterprise establishment (Pindyck and Rubinfeld 1991).

Conceptually, the mere existence of the option to migrate with good job prospects is, in the short run, expected to have a negative influence on local rural microenterprise establishment, which mostly provides jobs for individuals seeking to earn a living. A decision-maker may not be motivated by existing socio-economic conditions to stay and establish a local rural enterprise due to relatively more substantial influence of a potential migration option with good job prospects. The opportunity to migrate may make a decision-maker disinterested to consider other socio-economic variables favourable for microenterprise establishment. However, migration may not appeal equally to everyone due to differences in personal attributes which may align differently with pull factors in the destination region. The effect of the migration option may therefore vary among individuals. A utility function is therefore assumed to exist for them as rational decision-makers and with additive utilities accruing to them. Utilities are generated based on a combination of attributes at different levels, which form the basis for decision-making.

Stated preference methods are usually employed in the introduction of new schemes and use the individual respondents’ statements of their choices to estimate utility functions. Wardman (1988) notes that the principal drawback of stated preference is based on what people say they would do, as opposed to what people are observed to do. In contrast, revealed preferences involve inferring values from the choices of decision-makers made by observing their actual behaviour based on the prevailing factors. They miss the evaluation of decisions of individuals based on current scenarios. According to Kroes and Sheldon (1998), it can be challenging to obtain sufficient variation in the revealed preference data to examine all variables of interest. Therefore, revealed preference cannot be used in a direct way to evaluate choices under situations which do not yet exist.

OPEN IN VIEWER

Figure 2.

Source: The authors (2013).

Following the example of Dijkstra and Timmermans (1996) in the exploration of conjoint analysis in choice behaviour, a hypothetical scenario is presented in Figure 2 to explain the decision-making behaviour of the rural community dweller. The composite attributes (A) of each decision-maker combines with the options B₁₁ or B₁₀ representing presence and absence, respectively, of the migration factor. The scenarios also include socio-economic attributes (C) which are disaggregated into C1, C2, etc., and encompassing return on investment, infrastructural considerations and policy and institutional support. The decision-maker must examine the relevant scenarios and compare and rationalise them before deciding to establish a microenterprise. The option to migrate (B₁₁) in a scenario is expected to reduce the probability of a decision-maker to establish an enterprise within the locality. On the other hand, it may enhance the unfavourable socio-economic factors leading to a decision not to establish a local microenterprise. This could be a decision to out-migrate. Scenario 1 (SN₁) and Scenario 2 (SN₂) are different only for the migration option which exists for SN₂ but not SN₁. The probability of enterprise establishment for SN₁ is expected to be higher than SN₂. In this study, the choices of respondents are evaluated based on stated preferences rather than revealed preferences.

Research Methodology

The study employed a mixed method design that is both quantitative and qualitative. The quantitative aspect involved data collected through a cross-sectional survey of individuals from sampled households in selected rural communities in the municipality. The qualitative aspect employed focus group discussion (FGD), which complemented data from the survey. One FGD was held in each of the 20 study communities and served to explore in greater detail the knowledge and experiences of community dwellers and to understand further migration and microenterprise establishment issues that were raised.

The study population comprised individuals within the economically active age group of 15– 65 years who lived in rural localities in the Mfantsiman Municipality. A rural settlement has a total population of fewer than 5,000 persons (GSS 2013). Under Section 89 of The Children’s Act, 560 (1998), the minimum age for admission of a child to employment is 15 years. The total rural population in 2013 was estimated at 97,595. Of this, 53.8 per cent totalling 52,506 who fell within the economically active age group of 15–65 years constituted the study population. A total of 800 individuals who represent a cross-section of the study population were sampled from households within 20 selected rural communities in the municipality and interviewed.

A combination of stratified, systematic and simple random sampling procedures was used. First, the Mfantsiman Municipality was stratified based on the existing demarcation of the municipality into 256 enumeration areas (EA) during the 2010 Population and Housing Census (GSS 2013). Out of the 256 EAs, 134 were rural, consisting of 160 rural communities that were used for the study. Second, a systematic random sampling technique based on a sampling space of 8 was used to select 20 study communities from the 160 communities, constituting a sampling proportion of 12.5 per cent. This technique ensured that at least one rural community was selected from each area council. The houses were numbered and the number of households within each house listed. Third, the sampling of 40 households within each of the 20 communities using pseudo-random numbers. Finally, individuals were selected for an interview through simple random sampling (Borg and Gall 1989; Kumar 2005) using the list of economically active individuals within the sampled households. An initial sample size of 384 individuals was determined based on the sample size formula by Krejcie and Morgan (1970). However, to increase the probability of correctly rejecting the null hypothesis, a sample size of 800 was estimated from the study population of 52,506 at a significance level of 0.01, an error margin of 4.52 per cent and confidence level of 99 per cent. Calculation of power or sensitivity based on this yielded 86.5 per cent probability.

Focus groups were made up of individuals whose personal characteristics, previous or current occupations and dispositions enabled them to provide information on socio-economic conditions of the community for a detailed exploration and understanding of the themes under discussion. The first identifiable contacts in this study were the Assemblymen or Unit Committee members who were considered very important in the groups formed. They were purposively included because of their involvement in local development initiatives by the community, government and non-governmental organisations. The selection process of other members of the groups, employed snow-balling with interest in individuals whose occupations or backgrounds would enable detailed exploration and understanding of the themes under discussion (Bernard 2002). Youth below 40 years and women representatives were included to obtain their views on migration and enterprise establishment issues, and a deeper understanding of the facets that affected them.

Three field data collection instruments were used. The first was a household enumeration and respondent selection schedule used to build a sampling frame of eligible respondents between 15 and 65 years to be sampled for interview. It was in the form of a listing matrix that serially assigned a number to every house or compound in each study community. The house numbers of the selected houses were transferred unto the household enumeration and respondent selection schedule. The second instrument was an interview schedule for economically active individuals who were sampled using the first instrument. It was aimed at collecting data to analyse the attributes of respondents regarding their enterprise establishment decisions, based on 31 scenarios of combinations of the migration decision factor with other factors. Each scenario was explained to respondents to decide whether they would establish a rural enterprise or not. The third instrument was a FGD guide, which was used for discussions in each of the 20 study communities. An internal consistency technique was used by grouping questions on each of the attributes and sequencing them in such a way that doubtful answers could be validated by answers to subsequent questions.

Data collection instruments were used to train five assistants of the Ghana Statistical Service, Cape Coast, after which they were pretested at Mbroboto, a rural community in the Mfantsiman Municipality. Revision of the instruments after pretesting involved rephrasing and changing the order of questions to obtain better responses. Actual primary data collection included structured interviews of individuals in either English or the local dialect. The major challenge was the nature of the interviews, which required a lot of time with respondents to make them understand the objective of the exercise and obtain frank responses as possible. Five experienced field data collection assistants at the Ghana Statistical Service, Cape Coast, were engaged for primary data collection. Interviews of selected individuals were conducted in their homes at their convenience. Interviews and FGDs were mainly conducted on weekends and non-farm days. During the weekdays, interviews were conducted in the late afternoons.

Collation and editing of the interview schedules were performed by the entire field team. No interview schedule was rejected. Data processing involved data entry in a computer, cross-tabulations and estimation of analytical models. Computer-based software applications used to analyse quantitative data were Microsoft Excel, Statistical Product and Service Solutions (SPSS) version 16 and Stata version 12, while transcription and content analysis were applied to qualitative analyses based on common themes.

Based on the conceptual framework, 31 multi-attribute scenarios of the socio-economic attributes, denoted SN1 to SN31, were used to interview the 800 selected rural dwellers to predict the probability of enterprise establishment for each scenario. Implicitly, the individual respondent compared each of the defined scenarios with a scenario for which all socio-economic decision variables were absent in line with a study by Ryan et al. (1998). The dependant variable for the decision equation was a categorical variable with a binary response option, expressed as a probability and represented by a dummy. It took the value of 1 if the individual decided to set up a rural farm or non-farm enterprise and 0 if the individual decided not to set up an enterprise. The individual’s decision equation could thus be written as:

log ( P i / ( 1 − P i ) ) = α + β 1 E 1 + β 2 Mkt + β 3 Ins + β 4 Roi + β 5 Mg + ε i

(1)

The migration option (Mg) was the major variable of interest. Other decision variables, such as access to market (Mkt), access to institutional support services (Ins), availability of electricity (El) and perceived profitability or return on investment (Roi), were chosen based on reviewed literature and an initial analysis that examined various combinations of regressors. The attributes of a decision-maker, including age, sex and education, did not vary as he or she considered each scenario and therefore, were excluded from the ordinary logistic equation.

The constant α measures the logit estimate of the rural dweller, who would invest despite unfavourable factors. The β coefficients measure the predicted odds of microenterprise establishment, associated with a change from an unfavourable to favourable attribute or opportunity. The ordinary logit model was adopted against other binary choice models, because of the advantage of overcoming the difficulty of modelling a variable within a restricted range of 0–1 to predict the odds of an event within the real line. The a priori expectation was that there would be a negative relationship between the perception of a migration opportunity with good job prospects and the decision to establish a microenterprise. A negative intercept term is therefore expected to reflect a negative influence on the logit when the socio-economic attributes, such as access to market, access to institutional support services, availability of electricity and perceived profitability or return on investment, do not meet the enterprise establishment’s expectations of the decision-maker, or when the migration option is favourable to the rural dweller against microenterprise establishment.

The variance of the error term ε_i varies with the number of favourable decisions r_i. The error term was thus corrected for heteroscedasticity based on a recommendation by Cox (1970) and Domencich and McFadden (1975). The estimated equation was, therefore

Y = log r i + 1 2 / n i − r i + 1 2 = α + β 1 E 1 + β 2 MKt + β 3 Ins + β 4 Roi + β 5 Mg

(2)

Based on Equation (2), the predicted probabilities of the scenarios, with or without the migration variable, were compared to observe its influence on enterprise establishment decisions. Marginal analysis was carried out for the 31 scenarios of socio-economic determinants and compared. The marginal analysis gives an approximation of the probability of a positive decision from a unit change in the value of a decision factor. The rural dwellers’ attributes and socio-economic decision factors were analysed based on their ordinal rankings and the relative magnitudes of their marginal effects obtained from the decision equation.

To determine migration responsiveness among various age groups, the marginal effects of the migration option were compared using pooled data for age cohorts: 15–20 years, 21–30 years, 31–40 years, 41–50 years and 51–65 years. The use of the age cohorts was because of the possible masking effect associated with large age ranges. The significance of the marginal effect of the migration variable on microenterprise establishment decision-making was examined against the background that the higher their magnitude in the decision equation for that age group, the higher are the predicted odds of the rural dweller within that age group not establishing a microenterprise, and the higher is the potential to migrate. The effect size of the migration variable was analysed alongside the attributes of the decision-makers in the respective age category. Mann–Whitney U test was carried out to determine any differences in ranking of the various personal characteristics of respondents and socio-economic factors in microenterprise establishment.

Results and Discussion

This section begins with a subsection on the characterisation of respondents in the study area with a particular interest in their occupations and the incidence of migration. This is followed by the results of the predicted probability of microenterprise establishment in the presence of the migration option in line with the first objective of the study. The third subsection, which is related to the second objective of the study, presents the interactive effects of the migration option in microenterprise establishment in the presence of other decision factors.

Characterisation of Respondents in the Study Communities

Out of the 800 individuals interviewed, 449 (56.1%) were females and 351 (43.9%) were males. As shown in Table 1, 56.5 per cent of the respondents had no employable skills. Among those with employable skills, 31.5 per cent had vocational or technical skills and the rest who formed 12.5 per cent had acquired skills in agribusiness and other professions.

As shown in Table 2, the percentage of males with vocational and technical skills was higher than the females within each age group. Despite the higher educational attainment among males, both sexes undertook farming and trading as their primary occupation. The total number of people who were employed was 594. Out of this total, 322 (40.3%) were involved in agriculture while, 21.8 per cent were engaged in petty trading or food preparation for sale. The low scale at which these two occupations were undertaken did not require much formal education. About 25.8 per cent of the respondents who were not employed comprised 102 students, 40 apprentices and 64 others. Out of these 206 persons, 103 (50%) were women. Students and apprentices were not typically searching for jobs but occasionally engaged in income-earning activities such as farm labour, trading and burning wood for charcoal.

OPEN IN VIEWER

Table 1.

Professional or Employable Skills Acquired by Sex

Sex	Nil	Voc/Tech/Artisan	Tertiary	Agri-business	Other	Total	Percentage with Skills
Female	300	125	0	14	10	449	33.2
Male	152	127	6	16	50	351	56.7
Total	452	252	6	30	60	800	43.5
Percentage	56.5	31.5	0.8	3.8	7.5	–	–

Source: Field survey (2013).

OPEN IN VIEWER

Table 2.

Educational Level of Respondents by Age Group and Sex

Age Group	Sex	Nil	Pri	JHS	SHS	Voc/Tech	Tertiary	Total	% Educated
15–20	Total	2	24	110	48	0	0	184	98.9
21–30	Total	14	44	178	24	4	2	266	94.7
31–40	Total	36	42	70	12	0	6	166	78.3
41–50	Total	38	18	52	0	2	0	110	65.5
51–64	Total	18	10	40	0	4	2	74	75.7
15–40	Male	8	44	168	51	4	8	283	97.2
	Female	44	66	190	33	0	0	333	86.8
	Total	52	110	358	84	4	8	616	91.6
41–64	Male	14	8	38	0	6	2	68	79.4
	Female	42	20	54	0	0	0	116	63.8
	Total	56	28	92	0	6	2	184	69.6
15–64	Male	22	52	206	51	10	10	351	93.7
	Female	86	86	244	33	0	0	449	80.8
15–64	Total	108	138	450	84	10	10	800	86.5
	Percentage	13.5	17.25	56.25	10.5	1.25	1.25	–	–

Source: Field survey (2013).

The FGDs conducted at the 20 study communities showed that 60 per cent of migrants moved out once in a year, 2 years or more but stayed for periods less than a year. Those who left for short periods of usually less than 3 months were often students attending school outside the communities. Others visited friends or relatives in the cities. The main reason for out-migration was employment, which was ranked as first in importance based on the mean rankings in 16 communities. Apprenticeship or skill acquisition was the next in rank, followed by schooling and lastly, marriage. Job prospect, revealed as the main reason for out-migration, conforms to several studies (Chen 2014; De Brauw et al. 2002; Hashim 2005; Litchfield and Waddington 2003). The rankings were also consistent with findings by Ackah and Medvedev (2010) that the primary motivation for Ghanaian migrants is to find work, primarily in the manufacturing sector or trading, accounting for 49 per cent of all migrants who were 15 years old or more. The quest for education and marriage as motivation for migration were second and third reasons, which accounted for 16.5 per cent and 12 per cent of all migrants, respectively.

Effect of the Migration Option on Predicted Probability of Enterprise Establishment

The predicted probabilities of enterprise establishment, obtained from the pooled data of decision-makers for each of the 31 scenarios, are presented in Table 3. The predicted probabilities ranged from 0.367 to 0.89 for the various scenarios. The magnitudes of the predicted probabilities depict the relative preference of four socio-economic factor combinations in the presence or absence of the migration variable for microenterprise establishment.

OPEN IN VIEWER

Table 3.

Predicted Probabilities of Scenarios with or Without the Influence of a Migration Option in Enterprise Establishment

	Scenarios with Mg Included	Predicted Probability	Corresponding Scenario Without Mg	Predicted Probability
SN1	El, Mkt, Ins, Roi, Mg	0.881	SN6	0.890
SN2	Mkt, Ins, Roi, Mg	0.850	SN11	0.861
SN3	El, Ins, Roi, Mg	0.807	SN14	0.818
SN4	El, Mkt, Roi, Mg	0.793	SN12	0.805
SN7^	Ins, Roi, Mg	0.759	SN21	0.775
SN5	El, Mkt, Ins, Mg	0.726	SN10	0.743
SN15^	Mkt, Roi, Mg	0.746	SN26^	0.762
SN8	El, Roi, Mg	0.680	SN23	0.698
SN13^	Mkt, Ins, Mg	0.670	SN20	0.688
SN16^	El, Ins, Mg	0.595	SN25	0.615
SN17^	Roi, Mg	0.619	SN30	0.640
SN9^	El, Mkt, Mg	0.578	SN19^	0.598
SN22^	Ins, Mg	0.529	SN29^	0.551
SN24^	Mkt, Mg	0.511	SN28^	0.533
SN18^	El, Mg	0.431	SN27^	0.452
SN31	Mg	0.367	–	–

Source: Field survey (2013).

Notes: ^ Significant at 0.05 for attributes, mean difference = 0.017.

Out of the 31 scenarios, 24 scenarios which had predicted probabilities of 0.5 or higher are logical for enterprise intervention considerations. Data on 14 out of the 24 scenarios with predicted probabilities of 0.5 or higher did not provide a good fit to the ordinary logit model, implying that for these 14 scenarios, at least 50 per cent of the decision-makers in the study area would be motivated to consider setting up microenterprises irrespective of differences in their attributes.

The scenarios with a good fit to the logistic model are marked ‘^’ in Table 3. They indicate a statistically significant effect of differences in personal attributes on microenterprise establishment decisions. The scenarios that included the migration variable are compared with the predicted probability of the corresponding scenario without the migration variable in the fourth column. The mean difference in the predicted probability of enterprise establishment was 0.017, implying that the migration variable decreased the probability of enterprises establishment by 1.7 per cent. Most of this influence was due to people in the younger age groups. A t-test of difference in means with and without the migration variable showed significance at the 0.01 level and p-value of 0.000, implying that the migration option affected the decision behaviour of respondents. Related to this, Tutu (1995), Boakye-Yiadom and Mackay (2006), Yue et al. (2010) and Ackah and Medvedev (2010) indicated that younger persons had a greater tendency to migrate from rural to urban areas for job search, which adversely affected local enterprise establishment.

Interactive Effects of Migration Option on Microenterprise Establishment Decisions

Under the random effects logit, the marginal effects from two sets of conjoint models are presented in Table 4. The first is the main effects model, and the second is the model for the main effects of the socio-economic factors and their interactive effects with respondent’s characteristics.

Table 4 shows the marginal effects of the variables found to be significant: access to electricity (El), access to market (Mkt), institutional support (Ins), return on investment (Roi) and the migration option (Mg). These are based on 31 scenarios with attributes that were examined by 800 respondents in the decision to establish rural microenterprises. The interactive effects are presented only for personal attributes found to have a significant interactive effect on the migration variable. For both models, the estimated correlation of observations within respondents, Rho (ρ) is quite close. Similarly, the likelihood ratio test for ρ was significant at 1 per cent level for both models. The magnitudes of the marginal effects give a good approximation of the probability of a positive decision from a unit change in the value of the regressor. The marginal effects indicate that all the socio-economic factors were significant in affecting enterprise establishment decisions at the 0.01 level. The results show that the respondents consider the migration option to be least important as a decision variable.

OPEN IN VIEWER

Table 4.

Marginal Effects from the Random Effects Logit Model

Attribute Interaction	Main Effects and Interaction		Main Effects Only
	dy/dx	Std. Err.	Coef.	Std. Err.
Access to electricity (El)	0.110**	0.023	0.125**	0.009
Access to market (Mkt)	0.183**	0.026	0.237**	0.014
Institutional support (Ins)	0.230**	0.029	0.294**	0.017
Expected returns on investment (Roi)	0.409**	0.039	0.387**	0.027
Migration option (Mg)	−0.094**	0.022	−0.069**	0.007
Primary school education _Mg	0.040*	0.019	–	–
Good managerial and technical skills (Mts2) Mg	0.031**	0.011	–	–
Age4 (41–50)_Mg	0.083**	0.022	–	–
Age5 (51–65)__Mg	0.095**	0.025	–	–

Source: Field survey (2013).

Notes: * and ** Significant at 0.05 and 0.01, respectively.

The results of the main effects and interaction model show that the probability of enterprise establishment decreased by 9.4 per cent for a unit increase in the migration option. The interactive effects of a migration option with the attributes of respondents including, age, managerial and technical skills, and to some extent, educational attainment, reduce the effect of the migration variable in affecting the decision to establish a rural microenterprise. However, the interactive effect with the sex of respondents, their financial standing and level of marital or family obligations were not significant. This confirms the finding of Ackah and Medvedev (2010) that a person’s age, civil status and educational attainment are important determinants of the migration decision, while gender is not.

It can be inferred from Table 4 that the interactive effect of a migration option was significant and, therefore, important for respondents with primary school education. A 1 per cent change in the interactive effect due to a per cent increase in the migration option leads to 5.4 (−0.094 + 0.04) per cent decrease in the probability of rural microenterprise establishment compared to a 9.4 per cent decrease among those in the reference group with no education, as well as the other educational levels. This corresponds to the odds for enterprise establishment, which is 1.44 times more for the respondents with primary school education. The results are consistent with the finding by Weber et al. (2008) that more years of schooling will increase the probability of moving from a rural to a metropolitan county and therefore implies less tendency to establish a rural microenterprise.

For respondents with good managerial and technical skills, the probability of enterprise establishment from a 1 per cent increase in the migration option decreases by 6.3 per cent compared to 9.4 per cent for those with moderate or no skills. This is expected and confirms the finding of Boakye-Yiadom and Mackay (2006) that rural–urban migration mainly involves the young, versatile, skilled and better-educated members of the community. Among age groups, the probability of enterprise establishment due to an interactive effect from 1 per cent increase in the migration option decreases by 1.1 per cent and increases by 0.1 per cent, respectively, for those in the 41–50 and 51–65 age groups compared to 9.4 per cent decrease for the reference age group 15–20 years and the 21–40 age group. Respondents in the 41–65 age group are more likely to settle down to income-earning activities locally than migrating. This is in line with the finding of Yue et al. (2010) that the generation who are 35 years old and below tend to view migration as a form of investment with the accumulation of human capital. The insignificant difference in the marginal effect between the 21–30 year-olds and the 15–20 year-olds implies a similar interactive effect with the migration option.

Tables 5–8 present the results from the ordinary logit model disaggregated by various age cohorts. Table 5 shows the results of the effect of the decision variables for all respondents irrespective of age and sex. The magnitudes of the marginal effects indicate the relative influence on the probability of a positive decision to establish an enterprise as follows: Roi of 1.031, Ins of 0.661, Mkt of 0.59, El of 0.267. When evaluated irrespective of age, the insignificance of the migration variable is not surprising because of possible differences in importance that is attached to the migration option by the different age groups. Thus, the result may show a null effect. For this reason, the same regression was run based on age groups.

OPEN IN VIEWER

Table 5.

Marginal Effects of the Socio-economic Variables on Enterprise Establishment Decisions for All Respondents Irrespective of Age and Sex

Decision Factor	Marginal Effect	Std. Err.	T	Prob. > t	[95% Conf. Interval]
Access to electricity	0.267*	0.060	4.460	0.000	0.144	0.390
Access to market	0.590*	0.060	9.870	0.000	0.467	0.713
Institutional support	0.661*	0.060	11.060	0.000	0.538	0.784
Return on investment	1.031*	0.060	17.250	0.000	0.908	1.154
Migration opportunity	−0.086	0.060	−1.440	0.161	−0.209	0.037
Constant	−0.457*	0.080	−5.740	0.000	−0.621	−0.293

Source: Field survey (2013).

Notes: N = 31, F(5, 25) = 100.63, Prob. > F = 0.000, R² = 0.953, Adj. R² = 0.943.

* Significant at 0.01.

OPEN IN VIEWER

Table 6.

Marginal Effects of Socio-economic Factors on Enterprise Establishment Decisions by Respondents 15–20 Years of Age

Decision Factor	Marginal Effects	Std. Err.	T	Prob. > t	[95% Conf. Interval]
Access to electricity	0.386*	0.051	7.630	0.000	0.282	0.490
Access to market	0.602*	0.051	11.910	0.000	0.498	0.706
Institutional support	0.700*	0.051	13.850	0.000	0.596	0.805
Return on investment	1.223*	0.051	24.170	0.000	1.118	1.327
Migration option	−0.193*	0.051	−3.820	0.001	−0.298	−0.089
_cons	−1.296*	0.070	−18.490	0.000	−1.440	−1.151

Source: Field survey (2013).

Notes: N = 31, F (5, 25) = 186.36, Prob. > F = 0.000, R² = 0.974, Adj. R² = 0.97

* Significant at 0.01.

OPEN IN VIEWER

Table 7.

Marginal Effects of Socio-economic Factors on Enterprise Establishment Decisions by Respondents 21–30 Years of Age

Decision Factor	Marginal effects	Std. Err.	T	Prob. > t	[95% Conf. Interval]
Access to electricity	0.402**	0.100	4.040	0.000	0.197	0.608
Access to market	0.706**	0.100	7.090	0.000	0.501	0.912
Institutional support	0.788**	0.100	7.910	0.000	0.583	0.994
Return on investment	1.098**	0.100	11.010	0.000	0.892	1.303
Migration option	−0.200*	0.100	−2.010	0.050	−0.405	0.006
_cons	−1.138**	0.133	−8.570	0.000	−1.412	−0.865

Source: Field survey (2013).

Notes: N = 31, F(5, 25) = 47.57, Prob. > F = 0.000, R² = 0.90, Adj. R² = 0.88

* and ** Significant at 0.05 and 0.01, respectively.

OPEN IN VIEWER

Table 8.

Marginal Effects of Socio-economic Factors on Enterprise Establishment Decisions by Respondents 31–40 Years of Age

Decision Factor	Marginal effects	Std. Err.	T	Prob. > t	[95% Conf. Interval]
Access to electricity	0.239*	0.084	2.840	0.009	0.066	0.413
Access to market	0.532*	0.084	6.320	0.000	0.358	0.705
Institutional support	0.765*	0.084	9.090	0.000	0.591	0.938
Return on investment	0.856*	0.084	10.170	0.000	0.683	1.030
Migration option	−0.036	0.084	−0.430	0.674	−0.209	0.137
_cons	−0.601*	0.112	−5.350	0.000	−0.832	−0.370

Source: Field survey (2013).

Notes: N = 31, F(5, 25) = 43.4, Prob. > F = 0.000, R² = 0.896, Adj. R² = 0.877.

* Significant at 0.01.

Tables 6 and 7 show the results of the marginal effects of the decision variables for respondents in the 15–20 and 21–30 age groups. The results show statistically significant marginal estimates for the variables access to electricity and market, institutional support and return on investment at the 0.01 level with p-values of 0.000 in each case.

The marginal effect on the 15–20 and 21–30 age groups was significant for the migration variable at the 0.01 and 0.05 levels, respectively, with p-values of 0.001 and 0.050. The probability of establishing a rural microenterprise decreases by 19.3 and 20 per cent, respectively, for 1 per cent change in migration option for the two age groups. The relative magnitudes of the marginal effect for the two age groupings are consistent with their ranking in order of importance: return on investment, institutional support, access to market and electricity and migration option.

The higher marginal effect of the migration variable for the 21–30 age group than that of the 15–20 age group is due to the relatively higher desire to get a job within the age group of 21–30 years. This is because one might have completed schooling or less likely to be under the care of parents and, thus, ready to start life on their own. A migration decision to seek a job in this situation is more likely than that expected for the age group of 15–20 years. Thus, the marginal effect of the migration variable for the 15–20 age group was expected to be lesser than that of the 21–30 age group.

The results from Tables 5–7 are consistent with findings by Storey (1994), Boakye-Yiadom and Mackay (2006), Yue et. al. (2010) and Jasra et al. (2011). The insignificance of the migration variable and the low rank of the electricity variable are evident in these related studies. The FGDs revealed a high trend of migration among younger age groups. Discussants pointed mainly to migration of the young, educated people from the communities for different lengths of stay, primarily for job search. This was partly due to push a factor such as unavailable enterprise establishment options. On analysing return migrants in China, Yu (2013) similarly revealed that for almost all of those who were older than 40, the ageing factor was the primary determinant of returning to their home origins, especially females.

The marginal effect of the migration variable was not significant for the 31–40 age group, as shown in Table 8. The 31–40 age group was more likely to be encumbered with marital and family obligations than the higher age groups of 41–65 years and is more inclined to settle down to income-earning activities locally than migrating. The results of the 51–65 age group and those of the 41–50 age group showed the same trend as that for 31–40 age group, with the migration variable being statistically insignificant. A migration decision with job search among the 31–40, 41–50 and 51–65 age groups is, therefore, less likely compared to the 15–20 and 21–30 age groups. This is in line with findings of Boakye-Yiadom and Mackay (2006), Yue et al. (2010) that the youthful age groups are more responsive to the migration prospects and Ackah and Medvedev (2010) that the probability of being a migrant, rises until a person attains 36 years. The results in Tables 5–8 are consistent with those on the marginal effects of the decision variables from the random effects logit model in Table 4.

Conclusions and Recommendations

The article provides evidence of how the option to migrate affects the probability of rural microenterprise establishment among various age groups and the interactive effect of migration option on microenterprise establishment decisions. Relative importance is attached to personal attributes and socio-economic conditions as enterprise establishment decision factors. The migration option is the least influential factor affecting enterprise establishment decisions by both males and females among five decision factors in the order of importance: expected return on investment, institutional support services, access to market, access to electricity and migration option.

The option to migrate with job prospects significantly reduces the probability of enterprise establishment decisions in rural communities of the Mfantsiman Municipality by 1.7 per cent. The youthful age groups are more responsive to the migration prospects. The marginal effect of the migration variable is not significant among individuals who are 31–65 years old. The negative influence of the migration option on enterprise establishment is mainly among individuals who are 15–30 years old, but with a stronger influence among those who are 21–30 years old.

The probability of rural microenterprise establishment decreases when the migration option is considered alongside the attributes of respondents including age, managerial and technical skills as well as educational attainment. Thus, microenterprise establishment in rural communities is less likely with skilled and better educated members of the community. The probability of enterprise establishment with a migration option is much higher for those in the 41–65 age group than the 15– 40 age group. Migration decisions with job search are less likely in the 41–65 age group. The sex of rural dwellers in the Mfantsiman Municipality, their financial standing and marital or family obligations are not important in microenterprise establishment when considered alongside the option to migrate.

We recommend that development efforts by policymakers, private sector practitioners and authorities in the Mfantsiman Municipality, aimed at curtailing emigration through microenterprises interventions, should target potential beneficiaries by age, sex, educational attainment and technical skills. Particular attention should be paid to the utility derived by the youth between 15 and 30 years concerning socio-economic factors that influence microenterprise establishment decisions, namely access to electricity, institutional support services, market for outputs and returns on investment. Future studies that consider the migration option as a determinant of enterprise establishment should examine the motives and attributes of the migrants in destination regions.

Footnotes

Declaration of Conflicting Interests

Funding

The authors disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

ORCID iDs

William Yaw Amuesi Kodwiw

John Victor Mensah

References

1.

Ackah C. Medvedev D. . 2010. ‘Internal Migration in Ghana: Determinants and Welfare Impacts.’ International Journal of Social Economics 39 (10): 764–84. http://www.emeraldinsight.com/doi/pdfplus/10.1108/03068291211253386.

+ Show References

2.

Arthur D. D. 2004. The District Assemblies’ Common Fund and Development in Ghana. Lessons from the Mfantsiman District Assembly . Unpublished Master of Philosophy Thesis, University of Ghana, Legon.

Go to Reference

3.

Bernard H. R. 2002. Research Methods in Anthropology: Qualitative and Quantitative Methods . 3rd ed. Walnut Creek, CA: Alta Mira Press.

Go to Reference

4.

Boakye-Yiadom L. McKay A. . 2006. Migration Between Ghana’s Rural and Urban Areas: The Impact on Migrants’ Welfare . http://www.pegnet.ifw-kiel.de/global-images/grants/conference-2007/boakye-yiadom_mckay.pdf.

+ Show References

5.

Borg W. R. and Gall M. D. . 1989. Educational Research: An Introduction . 5th ed. New York, NY: Longman.

Go to Reference

6.

Chen D. 2014. Determinants of Migration: A Case Study of Ca Mau and Hue Provinces, Vietnam . Partial fulfilment of the requirements of the MA degree, Paper Presented to the Department of Economics, University of Ottawa, Ottawa.

+ Show References

7.

Cox D. R. 1970. Analysis of Binary Data . New York, NY: Chapman and Hall.

Go to Reference

8.

De Brauw A. Huang J. Rozelle S. Zhang L. Zhang Y. . 2002. ‘The Evolution of China’s Rural Labour Market During the Reforms.’ Journal of Comparative Economics 30: 353–529.

+ Show References

9.

Dijkstra J. Timmermans H. J. P. . 1996. Exploring the Possibilities of Conjoint Measurement as a Decision-making Tool for Virtual Way-finding Environments . Paper presented at CAADRIA’97. http://www.ds.arch.tue.nl/Research/publications/jan/Cdria97.pdf.

Go to Reference

10.

Domencich T. & McFadden D. L. . 1975. Urban Travel Demand: A Behavioural Analysis . Amsterdam: North-Holland Publishing Co.

Go to Reference

11.

Ewusi K. 1986. The Dimensions and Characteristics of Rural Poverty in Ghana . Technical Publication, No. 43. Institute of Statistical Social and Economic Research (ISSER), University of Ghana, Legon.

Go to Reference

12.

Ghana Statistical Service (GSS). 2005. 2000 Population and Housing Census: Central Region. Analysis of District Data and Implications for Planning . Accra: Asante & Hittscher Press Ltd.

Go to Reference

13.

Ghana Statistical Service. 2013. 2010 Population and Housing Census: Regional Analytical Report, Central Region . http://www.statsghana.gov.gh/publications.html.

+ Show References

14.

Hansson S. O. 1994. Decision Theory: A Brief Introduction . Stockholm: Department of Philosophy and the History of Technology, Royal Institute of Technology (KTH). http://people.kth.se/∼soh/decisiontheory.pdf.

Go to Reference

15.

Hashim I. M. 2005. Exploring the Linkages Between Children’s Independent Migration and Education: Evidence from Ghana . WP T12. Sussex Centre for Migration Research. http://www.migrationdrc.org/publications/working_papers/WP-T12.pdf.

Go to Reference

16.

Hoddinott J. 1994. ‘A Model of Migration and Remittances Applied to Western Kenya.’ Oxford Economic Papers 46 (3): 459–76.

Go to Reference

17.

Jasra J. M. Khan M. A. Hunjra A. I. Rehman R. A. U. Azam R. I. . 2011. ‘Determinants of Business Success of Small and Medium Enterprises.’ International Journal of Business and Social Science 2 (20): 274–80.

Go to Reference

18.

Krejcie R. V. Morgan D. W. . 1970. ‘Determining Sample Size for Research Activities.’ Educational and Psychological Measurement 30: 607–10.

Go to Reference

19.

Kroes E. P. Sheldon R. J. . 1998. Stated Preference Methods: An Introduction http://www.bath.ac.uk/e-journals/jtep/pdf/Volume_XX11_No_1_11-25.pdf.

Go to Reference

20.

Kumar R. 2005. Research Methodology: A Step-by-Step Guide for Beginners . http://www.sociology.kpi.ua/wp-content/uploads/2014/06/Ranjit_Kumar-Research_Methodology_A_Step-byStep_G.pdf.

Go to Reference

21.

Litchfield J. Waddington H. . 2003. Migration and Poverty in Ghana: Evidence from the Ghana Living Standards Survey . Sussex Migration Working Paper 10. Sussex, Centre for Migration Research.

+ Show References

22.

Mills J. S. 1906. Utilitarianism . Chicago, IL: University of Chicago Press.

Go to Reference

23.

Pindyck R. S. Rubinfeld D. L. . 1991. Econometric Models and Economic Forecast . Boston, MA: McGraw Hill Inc.

Go to Reference

24.

Ritchey P. N. 1976. ‘Explanations of Migration.’ Annual Review of Sociology 2: 363–404.

Go to Reference

25.

Ryan M. McIntosh E. Shackley P. . 1998. ‘Methodological Issues in the Application of Conjoint Analysis in Health Care.’ Health Economics 7 (4): 373–78.

Go to Reference

26.

Stark O. 1980. ‘On the Role of Urban-to-Rural Remittances in Rural Development.’ Journal of Development Studies 16 (3): 369–74.

Go to Reference

27.

Storey D. J. 1994. Understanding the Small Business Sector . London: Routledge.

Go to Reference

28.

Todaro M. P. 1969. ‘A Model of Labour Migration and Urban Unemployment in Less Developed Countries.’ American Economic Review 59: 138–49.

Go to Reference

29.

Tutu K. 1995. ‘Determinants of Internal Migration.’ In Migration Research Study in Ghana , edited by Twum-Baah K. A. , Nabila J. S. , and Aryee A. F. , 97–113. Accra: Ghana Statistical Service.

Go to Reference

30.

Twumasi-Ankrah K. 1995. ‘Rural-urban Migration and Socio-economic Development in Ghana: Some Discussions.’ Journal of Social Development in Africa 10 (2): 13–22.

Go to Reference

31.

Wardman M. 1988. A Comparison of Revealed Preference and Stated Preference of Travel Behaviour . http://www.bath.ac.uk/e-journals/jtep/pdf/VolumeXX11No_1_71-91.pdf.

Go to Reference

32.

Weber B. Marre A. Fisher M. Gibbs J. Cromartie J. . 2008. Education’s Effect on Poverty: The Role of Migration and Labour Markets (Working Paper RSP 07-01). http://ruralstudies.oregonstate.edu/sites/default/files/pub/pdf/rsp_reports/rsp07-01.pdf.

Go to Reference

33.

Wirawan I. B. 2008. Circular Migration of Female Workers to Abroad: A Study on Migration Decision-making Processes by Females from East Java Villages . http://www.damandiri.or.id/file/idabaguswirawanunairabstrak.pdf.

Go to Reference

34.

Yu L. 2013. Labour Market Outcomes, Migration Intentions of Rural-urban Migrants and Return Migration in China . https://www.uleth.ca/dspace/bitstream/handle/10133/3340/yu,%20li.pdf?sequence=1.

Go to Reference

35.

Yue Z. Li S. Feldman M. W. Du H. . 2010. ‘Floating Choices: A Generational Perspective on Intentions of Rural-urban Migrants in China.’ Environment and Planning A 42 (3): 545–62.

+ Show References