What Causes Industry Agglomeration,2010
内容简报
构建coagglomeration indices,将这一指数与产业关联、劳动力池、知识溢出和自然禀赋联系起来,考察各个因素对聚集的影响。发现自然禀赋解释了产业聚集的40%,马歇尔的理论也都得到了验证。
精彩部分
考虑的是产业间的协同聚集,而不是产业内的聚集。
产业的聚集,已经超越了地理空间。其布局是基于全球战略和全球资源的。
The EG coagglomeration index of industry i and j is
γ_ij^c=(∑_(m=1)^M▒〖(s_mi-x_m)(s_mj-x_m)〗)/(1-∑_(m=1)^M▒x_m^2 )
where m indexes geographic areas. s_mi is the share of industry i’s employment contained in area m. x_m measures the aggregate size of area m, which we model as the mean employment share in the region across manufacturing industries.
DO indices (就业人口加权)
K ̂_ij^Emp (d)=1/(h∑_(r=1)^(n_i)▒∑_(s=1)^(n_j)▒〖e(r)e(s)〗) ∑_(r=1)^(n_i)▒∑_(s=1)^(n_j)▒〖e(r)e(s)f((d-d_(r,s))/h)〗
where d_(r,s) is the Euclidean distance between plants r and s, f is a Gaussian kernel density function with bandwidth h, n_i and n_j are the number of plants in industries i and j, respectively.
具体见DO-指数一文的1095页,公式(6)。
四大指标
(1)产业间的投入比例和产出比例:表达货物的运输成本
we define undirectional versions of the input and output variables by Inputij = max {Inputi←j, Inputj←i}and Outputij =max {Outputi→j, Outputj→i}. We also define a combined
InputOutputij= max {Inputij, Outputij}.
(2)劳动力池
the 1987 National Industrial-Occupation Employment Matrix (NIOEM) matrix provides industry level employment in 277 occupations, and we define Shareio as the fraction of industry i’s employment in occupation o. We measure the similarity of employments in industries i and j through the correlation of Shareio and Sharejo across occupations.
(3)知识溢出
We base our metrics of information flows on patents and research and development (R&D), which reflect only the highest level of information flows, rather than worker level spillovers.
(4)自然禀赋
This methodology follows Ellison and Glaeser (1999), who model 16 state level characteristics that afford natural advantages in terms of natural resources, transportation costs, and labor inputs. Combining these cost differences with each industry’s intensity of factor use, Ellison and Glaeser (1999)estimate a spatial distribution of manufacturing activity by industry that would be expected due to cost differences alone.
We employ these expected spatial distributions of industries across states to calculate expected coagglomeration levels CoaggijnA for industry pairs. Separate expected coagglomerations due to natural advantages are constructed for the EG and DO metrics. These measures simply substitute the predicted spatial employments by industry into the EG and DO formulas outlined in Section I. Essentially, this approach measures how coagglomerated the two industries would be if their locations were determined entirely by the interactions of industry characteristics and local characteristics. The DO metric again requires some slight modifications, which we document in the online Appendix. The pairwise correlation between expected and actual coagglomeration using this technique is 0.2 and 0.4 for the EG and DO techniques, respectively.
|