It is assumed that the (spinning) workers move from j ken to i ken in numbers (n^i_j), following the relation described below. log nn^i_j=a_0+a_1(w^p_j/wi)+a_2A_j+a_3φ(τn^i_j, θn^i_j), where ∂(log nn^i_j)/∂(w^p_j/w_i)<0, ∂(log nn^i_j/∂A_i<O, ∂(log nn^i_j)/∂ψ>0. A_j stands for average agricultural incomes per family and (w^p_j/w_i) for the wage differentials between regions (ken's). [w^p_j is calculated as follows: w^p_j=(Σ__in^i_jw_i-n^p_jw_p)/(Σ__in^i_j-n^p_j), where suffix p denotes "excluding any specific i."] The term ψ is a shift variable in the above job application equation and supposed further to be a function of τ^i_j, the familiarity of the applicants in j ken with i ken, and also of θ^i_j, the order of hiring preference to the workers from j ken held by the spinners in i ken. In the proceeding analysis (Obi and Nishikawa, Mita Annals of Economics vol. 4. 1960) the ψ-effect was conveniently called as "distance effect," and identified with the accumulative numbers of in- and out-flows between i and j ken, N^i_j(t)=Σ^^<t-1>__<t=0>n^i_j(t)[o stands for any basic year]. The N^i_j may be interpreted as a crude index of past history on labor movement between i and j ken. The introduction of N^i_j instead of ψ presents us a sufficiently high and significant multiple correlation; however there still remain some gaps between our understandings and the actual course of events. It is the purpose of this second report to fill these gaps. It is supposed that ψ-ffect is composed of f_i and f_t, which are unspecified factors and represent inter-regional and inter-temporal effect, respectively. logn^i_j=f^i_j+f^i_t+a_1(w^p_j/w_i)+a_2A_j ; i-e., a_3ψ=f^i_j+f^i_t. The f^i_j-s and f^i_i-s are estimated from the inflows from j to i. As for i, Aichi (23), Hyogo (28) and Okayama (33) are selected. Our subsequent analysis is to correlate any variables X_q(_i) with f^i_j. The f^i_j-s are 1. positively correlated with M_i, total outflows from j ken, 2. positively correlated with M^i_j, outflows from j to i ken, 3. not significantly correlated with(M_j/P_j), the ratio of outflows to population. The variables M_j, M^i_j and (M_j/P_j) are considered to be alternative Indices of psychological tendencies to move out from the native area. The f^i_j-s are 4. positively correlated with R_j(t), silk reeling outflows from j ken at year t, 5. positively correlated with S_j(t), numbers of silk reeling workers employed in j ken at year t, 6. positively correlated with ΔR_j (5-10) and ΔS_j (5-10) ; these are decrements of R_j and S_j from S. 5 (1930) to S. 10 (1935). It is inferred according to Findings 4~6 that the decreases of job opportunity in silk reeling accerelate the outflows to spinning mills. In this context we are remined that the geographical distributions of these two industries are considerably different. The silk reeling are scattered widely ail over the country and spinning mills concentrated in 4 areas along the coast of Pacific Ocean. This characterizes the scopes of labor mobility of the workers in these industries ; that is, the workers in silk reeling moved narrowly, while the workers in spinning mills moved far more long distance. Findings 4~6 are clearly observed only for inflows to Aichi (i=23), but not for Hyogo (i=28) and Okayama (i=33). This is because both the two industries are highly mixed in Tokai and Tosan District, while in the other areas, the relative weight of silk reeling workers are too low to result in such high and significant correlations. Besides what mentioned above, we have reviewed some case-studies on labor market organization, undertaken by emploment offices nearly at the period in consideration.
|