Factors Affecting Labor Participation: Results and Discussion of Multinomial Logit Model

This study has considered factors that determine household decision to participate in rural labor market in Northwest Pakistan using multinomial logit model. The results for the multinomial logit model are presented in terms of logit coefficients and odd ratio¹⁰⁹ in Table 38 and Table 39 respectively. Since direct interpretation of coefficient (Table 38) is difficult for the estimates from multinomial logit model, the odd ratio (risk ratio) is used to describe the impact of variables on employment choices (Table 39). We use non-farm informal sector as the comparison group, which happens to be the dominant (33%) employment strategy in the study area. Stata (ver. 9.2) is used to estimate the multinomial logit model. The chi-square for the model is signifi-cant at 1 percent level, and more importantly, pseudo R² explains over 28 percent of the variance among the groups.

109 It makes more sense to report the results of multinomial logit model in odd ratios than marginal effects as the MNLT is a multi-equation model and sometime the coefficients switches sign between raw coffiicients and its counterparting marginal effects (see for details on these issues, G^REENE, 2003; L^ONG & F^REESE, 2001).

Table 38. Multinomial Logit Regression for Occupational Categories

Note: Those engaged in informal wage activities serve as base group.

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

1) Village Yousafkhel served as reference village.

Source: Basic survey carried out in six villages of Northwest Pakistan, 2004-05.

Table 39. Odd Ratios for Multinomial Logit Regression for Household Occupations

Note: The coefficients for each occupations measure the odd ratio relative to informal wage activities.

Standard errors in parentheses

1) Village Yousafkhel served as reference village.

*** p<0.01, ** p<0.05, * p<0.1

Source: Basic survey carried out in six villages of Northwest Pakistan, 2004-05.

8.4.1 Role of Human Capital in Labor Participation

Our results support the human capital hypothesis. The year of education variable is significant and positive in almost all comparisons (OR=1.10-1.34 p<0.01). Additional years of education prompt the respondents into occupations other than non-farm informal sector, ceteris paribus (see Table 39). Nevertheless, the size of education coefficient is highest (odd ratio in

range of 32-34%) for farmers with non-agricultural jobs (mixed farmers) and formal sector earners. Results are in congruence with earlier findings(SCHMIDT & STRAUSS, 1975; AMELIE &

ZIMMERMANN,2004).

Age of the household head as proxy for experience, turn out to positive and significant for nearly all categories (OR=1.11-1.28 p<0.01-0.1). Contrary to expectations, the likelihood of sticking to pure farming and own businesses as compared to informal wage activity is not generally influ-enced by age of the household head¹¹⁰. Older household heads are expected to be involved in occupations other than informal wage earning since they have higher accumulated experience.

The negative sign of age square variable indicates the nonlinearity in age (Table 38). There are a few possible reasons for this phenomenon:

a) Firstly, it was observed in the study area that majority of those employed in the formal sector were in lower cadre jobs, increasing the likelihood for their early retirement.

b) Secondly, the introduction of ‘Golden Handshake Scheme’ in Pakistan under the influ-ence of international financial institutions during the structural adjustment programme era motivated a large number of formal sector workers to retire early for getting benefits;

and

c) Lastly, the joint family system is more common among farming households in rural Paki-stan and so elders don’t work so often. Similar results were reported by ARIF et al., 2000 in their study in Pakistan.

As expected the household size turn out to be positive and significant for majority of the con-strasting occupational categories (OR=1.04-1.11 p<0.01-0.1). A large household because of dependency problems, make it necessary on part of the household to diversify their income sources and strive for occupations that are more rewarding than informal wage jobs. In addition, in a Pashtun society, household composition also affects the types of employment its members enter, as well as how or by whom decisions are taken (ALBRECHT, 1976; MANIG, 1991). The variable on working members in a household showed mixed picture. It turned out to be negative and significant (OR=0.34-0.43 p<0.01) for the pure farmer and pure tenant categories relative to reference group. However, the reverse (OR=1.41-1.50 p<0.01) was true for those enaged in mixed farming and mixed tenancy compared to informal wage jobs. This reveals that

110 Younger household members with better education could have a better paid job, however, we considered only the age of the household head. The study do acknowledge the limitations of these results.

as a strategy of survival, majority of the household members within the informal wage earning household have to work.

8.4.2 Role of Household Assets in Labor Participation

As expected ownership of land and livestock, stimulate households to farming (Table 39).

Except those in own businesses and formal sector jobs, the probability (odds) increases signifi-cantly for livestock ownership for the rest of alternative occupational categories (OR=2.17-8.12 p<0.01). Turning to land ownership, Table 39 indicates that in odds, it matters almost 2-5 times more for the households to be involved in pure or mixed farming than in informal wage activities (OR=1.71-4.61 p<0.01). Wealth, defined here as household per capita income is expected to matter a lot in the likelihood of pursuing occupations other than informal work in the study area.

However, because of possible endogeneity problems, the variables on income and credit were dropped from the model.

8.4.3 Role of Geographical Location and Infrastructure in Labor Participation

Finally comparing the six villages in term of labor market, we find that relative to Yousafkhel, the probability increases for the rest of villages, that households pursue businesses or work as tenants than informal wage works (Table 39). However, for the rest of occupation categories, except Mushtarzai majority of the results are negative and significant. Tenancy is strong in Kochian (OR=2.68-15.0 p<0.01), where the landlord rent-out their land in small parcels on share cropping based on 50-50 share in produce.

Working as businessmen or independent trader is normally adopted by those (village Kukar &

Dalazak) around the provincial capital Peshawar, or near the main road Charsadda highway (village Kochian). This on one hand allows for easy commuting and on other hand satisfies the service demands of the urban dwellers in the provincial capital Peshawar as well as the city of Charsadda. However, in case of Mushtarzai, the demand for business related activities is mostly local driven where the remittance from abroad played a major role in setting up or initiating self employment.

Being a resident of Dalazak, Gulbela, Kochian, and Kukar decreases the likelihood of involve-ment in formal sector jobs (regulated jobs in governinvolve-ment and private sectors) and mixed farming as opposed to the base category. Nevertheless, the coefficient for Mushtarzai is statistically positive (OR=1.51-1.85 p<0.05-0.1) for these occupational categories. The less probability for formal sector jobs in the villages near the provincial capital Peshawar points towards two main realities in the area:

a) It shows an emerging positive tendency on part of the dwellers in the tribal village Yousafkhel with its rainfed nature to find their livelihood in more secure, permanent, and high return occupations. Due to their agro-climatic conditions, the working adults in Yousafkhel and also in Mushtarzai either work in gulf countries or moved for perma-nent jobs to big metropolis in Pakistan like Karachi, Lahore, and Islambad.

b) The dwellers of Yousafkhel and Mushtarzai being located at the tribal belt have more chances to qualify for working in armed forces in Pakistan and border patrol jobs due to the quota system for FATA.