产业结构和经济稳定【外文翻译】

《产业结构和经济稳定【外文翻译】》由会员分享，可在线阅读，更多相关《产业结构和经济稳定【外文翻译】（11页珍藏版）》请在装配图网上搜索。

1、本科毕业论文外文翻译外文题目： Industrial structure and economic stability 出 处： Applied economic letters 作 者： Sherrill Shaffer 原 文：Motivated by prior results predicting contrasting linkages between industrial structure and economic stability, we present exploratory empirical evidence on this important issue. Consi

2、stent with the turnover hypothesis, we find that employment grew more steadily where business establishments in all sectors were larger, suggesting an offsetting benefit to the first-moment costs of establishment size identified by previous research. Consistent with the job-matching hypothesis, we f

3、ind that employment grew more steadily where more establishments per capita operated in all sectors. Similar but less consistent results were also found regarding the stability of income growth.I. Introduction and BackgroundThe fundamental importance of economic performance has spawned an extensive

4、literature on the empirical determinants of economic growth (see Rajan and Zingales, 1998; Levine et al., 2000 for prominent examples). One recent discovery is that communities or local regions tend to experience more rapid growth of income or employment where businesses are smaller (Shaffer, 2002,

5、2006a, b). A separate strand of research, meanwhile, has demonstrated the importance of studying the volatility of growth rates rather than merely their means (Ramey and Ramey, 1995; Kurz, 2004; Ismihan, 2005; Bekaert et al., 2006). It is relevant in this regard that several studies (Davis and Halti

6、wanger, 1992; Rob, 1995; Davis et al., 1996) have found that jobs at smaller firms tend to be less permanent than at larger firms; while Ilmakunnas et al. (2005) have suggested that such turnover may actually be one reason for faster productivity growth, an unexplored implication is that the volatil

7、ity of employment and income may be higher where businesses are smaller (the turnover hypothesis).Conversely, heterogeneity among decision-makers due to human fallibility can result in greater variability of economic performance in more centralized decision processes (Sah, 1991; Sah and Stiglitz, 19

8、91; Almeida and Ferreira, 2002), possibly suggesting that a local economy may exhibit greater stability where the average firms are smaller or more numerous. A similar outcome is predicted to the extent that (a) centralized decision processes are less successful at managing conflict and (b) distribu

9、tional conflicts impair efficient adjustments to exogenous shocks (Rodrik, 1999; Almeida and Ferreira, 2002).In view of these contrasting considerations, an important empirical question is whether establishment size and other measures of industrial structure may be systematically associated with sec

10、ond moment measures of economic performance.This article accordingly presents preliminary evidence of linkages between selected measures of industrial structure and the volatility of local income and employment growth rates. We use three measures of industrial structure, each distinguished by broad

11、sector. As in Shaffer (2002), we measure average establishment size by number of employees and, alternatively, by dollars of value added, shipments or receipts. We also look at establishments per capita, motivated by two opposing considerations. Finding a new job should be easier in a market with mo

12、re employers in a given sector, leading to more stable levels or growth rates of income and employment (the job-matching hypothesis). But, ceteris paribus, smaller firms will tend to permit the coexistence of larger numbers of firms, in which case the documented employment turnover at smaller firms

13、(discussed above) will tend to offset the stabilizing benefit of more numerous firms. Each of our measures of industrial structure is compiled separately for the manufacturing, wholesale, retail and service sectors.We relate these measures of structure to two second-moment measures of economic perfo

14、rmance, the SD of annual real per capita income growth rates and the SD of the annual growth rate of total establishment employment. The results contribute to two separate strands of the literature, on empirical covariates with growth volatility and on macroeconomic effects of establishment size and

15、 other measures of industrial structure. We find for all sectors that larger establishments and more establishments per capita are associated with more stable employment growth rates, consistent with the turnover and job-matching hypotheses. The same linkages are found for some but not all of the se

16、ctors with regard to the volatility of growth rates in real per capita income.The next section introduces the empirical model and the sample. Section III presents the results, while Section IV concludes.II. The Model and SampleWe embed our key variables in a standard linear empirical growth equation

17、, (1)Where Y is a measure of economic performance as discussed above, is an estimated intercept term, s is a measure of industrial structure as discussed above, x is a vector of control variables discussed below，andare estimated coefficients and is a stochastic error term. As in Bekaert et al. (2006

18、), our SDs of economic growth rates are measured over a 5-year period. As in prior studies of economic growth, the control vector includes the natural logarithm of population, the density of population per square mile of land area, a measure of education and initial median household income.Populatio

19、n is a measure of market size as in Cetorelli and Gambera (2001). It is also similar to the total labor force variable used in Ohuallachain and Satterthwaite (1992) and can be interpreted as measuring urbanization economies. If job-matching occurs more quickly or efficiently in more populous areas,

20、then the estimated coefficient on this variable should be negative (another implication of the job-matching hypothesis introduced above).Population density has been found significantly related to several first-moment measures of economic performance, possibly due to scale effects or to superior matc

21、hing between firms and workers in denser markets (Ciccone and Hall, 1996; Andersson et al., 2004; Carlino et al., 2007; Strumsky et al., 2005). If these benefits influence economic stability, as one might expect, then the estimated coefficient on population density should be negative in our model.Ed

22、ucation is measured as the percentage of population aged 25 and over who have completed high school, and reflects the accumulated level of human capital. It is similar to measures used in previous studies of economic growth such as Rajan and Zingales (1998), Levine et al. (2000) and Cetorelli and Ga

23、mbera (2001), with theoretical linkages to average growth rates explored by Teles (2005).A related measure of education has been used in at least one study of growth volatility (Bekaert et al., 2006). In addition, education was found to be positively associated with sectoral employment growth in US

24、metropolitan areas by O huallachain and Satterthwaite (1992). If education contributes to economic stability as well, its estimated coefficient in our regressions should be negative.Initial median household income can reflect a convergence effect in first-moment measures of economic performance, as

25、noted by Barro and Sala-i-Martin (1992). The same logic would not apply to second-moment measures of economic performance, rendering the sign and significance of the estimated coefficient on this variable an open empirical question. However, a similar variable (initial per capita GDP) has been used

26、in at least one prior study of growth volatility (Bekaert et al., 2006).Table 1 summarizes the data. Our sample comprises more than 2000 US nonmetropolitan counties, measuring economic performance during 19911995 and structure measures as of 1987. Metropolitan areas are excluded because their border

27、s are generally not coterminous with individual counties, confounding measurement problems for variables drawn from county-level data. Though not reported in the table, the pairwise correlation coefficients between average establishment size and establishments per capita ranged between 0.21 and 0.18

28、, and were just 0.07 in the wholesale sector and 0.07 in the manufacturing sector. These small and highly variable correlations indicate that the two categories of structure variables reflect statistically separate dimensions of industrial structure in our sample.The selection of performance data as

29、 of several years following the structure data helps to reduce the likelihood of reverse causality although, common to all empirical growth studies, causality cannot be definitively established. This lag structure also minimizes the potential for endogeneity bias, as the regressors are predetermined

30、.III. ResultsTable 2 reports the regression estimates for the SD of real per capita income growth rates, while Table 3 reports estimates for the SD of employment growth rates. As industrial structure is measured in three ways for each of four sectors, each table reports 12 regressions. Due to missin

31、g or zero establishment data for a few counties in each sector, the various regressions utilize slightly differing numbers of observations, as reported in the tables. The significance levels are computed from SEs corrected for heteroscedasticity.In Table 2, the various structure measures are statist

32、ically significant in eight of the 12 regressions. Per capita income is found to grow more steadily where manufacturing, retail and wholesale establishments employ more workers, or where wholesale establishments are larger as measured by value of annual shipments. These findings are consistent with

33、the turnover hypothesis discussed above.Conversely, steadier growth is seen where wholesale establishments employ fewer workers or in counties with fewer wholesale establishments per capita or more manufacturing establishments per capita. The contrasting estimates for the wholesale sector may reflec

34、t nonlinearities in the underlying relationships, an issue beyond the scope of this study. These findings indicate that industrial structure is systematically related to the volatility of income growth rates, in ways that are largely consistent but that do vary somewhat across major sectors.The fit

35、of the equations in Table 2 is modest, with adjusted R2 ranging from 0.3 to nearly 0.5. The control variables are mostly significant at the 0.01 level. Initial median household income is never significant, in contrast to the significance of initial per capita GDP found in several cross-country regre

36、ssions on consumption growth volatility by Bekaert et al. (2006). Income is found to grow more steadily in more populous counties. Population density and education are found to contribute to higher volatility of per capita income growth rates, though education loses its significance when controlling

37、 for wholesale establishments per capita. The sign of the education variable contrasts with that found in cross-country specifications by Bekaert et al.(2006).In Table 3, the various structure measures are again statistically significant in eight of the 12 regressions. Employment is seen to grow mor

38、e steadily where there are more establishments per capita in any of the four sectors, consistent with the job-matching hypothesis. In addition, steadier employment growth occurred where wholesale or service establishments employ more workers, or where retail or service establishments have a larger a

39、verage dollar volume of business, consistent with the turnover hypothesis. These latter findings indicate that the benefits of job longevity at larger firms more than offset the volatility theoretically predicted by more centralized decision-making, as discussed above.The fit of these equations is l

40、ess tight than in Table 2, with adjusted R2 around 0.1. Population density is never significant but the other control variables are consistently significant. Employment grows more steadily in more populous counties, where education is higher, or where initial median household income is lower. The po

41、pulation result is consistent with the job-matching hypothesis discussed above, and the education and initial income results match those found in cross-country regressions for consumption growth volatility by Bekaert et al. (2006). As in Table 2, education loses its significance when controlling for

42、 wholesale establishments per capita.IV. ConclusionMotivated by prior results suggesting contrasting and important predictions relating industrial structure to economic stability, we have explored empirical linkages between multiple sector-specific structure measures and the volatility of subsequent

43、 growth rates in real per capita income and employment. Significant associations were found in two-thirds of the specifications, controlling for a standard set of environmental variables. Overall, the turnover hypothesis was found to dominate the contrasting predictions of inefficient centralized de

44、cision-making, as employment and income both tended to grow more steadily where establishments were larger. These findings suggest an offsetting benefit to the first-moment costs of establishment size identified by Shaffer (2002, 2006a, b).The job-matching hypothesis is consistent with steadier grow

45、th where more establishments per capita operate, as found for all sectors in the employment regressions and for the manufacturing sector in the income regressions. Contrasting results were found in the income regressions for wholesale establishments measured by employment size and for wholesale esta

46、blishments per capita. Although our multi-year lag structure reduces the likelihood of reverse causality between industrial structure and growth volatility, causality cannot be definitively established, a shortcoming common to all empirical growth studies. Consequently, policy inferences should be d

47、rawn with caution, pending additional research on the precise mechanisms underlying the observed patterns.译 文：产业结构和经济稳定 谢里尔谢弗 经济及金融系电子邮箱：shafferuwyo.edu 通过预测产业结构和经济稳定之间的联系，在事先对比结果的启发下，我们提出探索这一重要问题的证据。在符合营业额假设的前提下，我们发现更稳定的就业增长在各行业机构均较大，这意味着一个抵消造福于编制第一时刻由以前的研究中发现的成本。这与工作匹配假说相一致，即就业增长在经营各行业的场所更加稳定。与此相类

48、似，但不太一致的结果也发现关于收入增长的稳定性。一、引言和背景经济运行的重要性催生了大量关于影响经济增长的实证因素的文献（见拉詹和辛格勒，1998；莱文等人，突出的例子为2000）。最近的一项研究发现，社区或局部地区往往经历着更快速的收入或就业增长，尤其是在小规模企业（谢弗，2002，2006a，b）。一个单独的研究链，它同时展示了学习增长率波动的重要性，而不仅仅是他们的手段（拉米和拉米，1995；库尔兹，2004；Ismihan，2005；贝卡尔特等，2006）。正是在这方面的几项有关的研究（戴维斯和黑尔蒂旺热，1992；罗布，1995；戴维斯等人，1996）发现，在大公司工作往往比在小企业

49、更持久；而Ilmakunnas等（2005）建议，这类营业额实际上可能是一个更快的生产率增长的原因，一个未开发的含义是就业和收入的波动性可能会导致更高的营业额在规模较小的地方企业中（即营业额假说）。相反，决策者由于人类的不可靠而导致经济的差异性更加集中地表现在决策的过程中（萨，1991；Sah和斯蒂格利茨，1991；阿尔梅达和费雷拉，2002）。这意味着，在一个企业平均规模较小或企业较多的地方，它的经济可能会更加稳定。一个在预计范围内的类似结果：（一）集中决策过程不太善于管理冲突和成功。（二）分配的冲突损害效率调整（罗德瑞克，1999；阿尔梅达和费雷拉，2002）外生冲击。基于这些对比的考虑，

50、一个重要的经验问题是，是否建立规模和产业结构等措施可能是系统地与经济表现的二阶矩措施相关。本文据此提出了初步的证据，关于选择工业结构和地方收入和就业增长率的波动性的相应措施之间的联系。我们使用的三种产业结构措施受到每个部门的广泛尊敬。正如谢弗（2002），我们衡量平均的建立规模，通过员工人数、创造美元的附加值、发货或收据。我们也期待由两个对立因素推动的人均机构。寻找一份新的在某一有很多雇主的部门的工作应该更容易，它导致更多稳定水平的收入和就业增长率（即“就业选配假说”）。但是，其他条件不变，小企业往往会允许更多数量的企业并存，在这种情况下，在小企业就业（如上所述）记录的营业额往往会抵消许多公司

51、更加稳定的利益。我们产业结构的每一个措施都被单独编译为制造业，批发，零售和服务行业。我们称这些结构措施为经济表现的两个二阶矩措施，即每年的人均实际收入增长率统计处和编制总额的年均增长速度的SD就业这些措施。这个结果导致了两个单独的文献股，经济增长波动的实证协变量和建立规模与其他产业结构措施对宏观经济的影响。我们发现所有规模较大的企业和人均较多的部门都与稳定的就业增长速度相关联，符合营业额和就业选配假设。同样的联系是找到了，但并不带来所有部门人均实际收入增长速度的波动。下一节介绍了经验模型和样本。第三节介绍结果，而第四部分为结论。二、模型及样本我们将关键变量嵌入一个标准的线性增长方程， (1)其

52、中Y是衡量经济表现，如上所述，是一个估计的截距项，S是产业结构衡量标准，因为上面的讨论，X是下面讨论的控制变量向量，是估计系数，是随机误差项。如贝卡尔特等（2006），我们的经济增长率已经被测量了5年的时间。正如之前的经济增长的研究，控制矢量包括人口的自然对数，平均土地面积的人口密度，教育和初始家庭收入中位数的衡量。人口是市场规模的测量措施在Cetorelli和Gambera（2001）。这也类似于总劳动人口的变量在Ohuallachain和Satterthwaite（1992）的使用，它可作为衡量城市化的经济解释。如果就业选配起来更快或者更有效地发生在人口较多的地区，那么对这个变量的估计系数

53、应为负（另一个上面介绍的就业选配假说的含义）。人口密度已被发现与经济表现的几个第一时刻措施显著相关，可能是由于规模效应，或企业和工人之间的匹配，优越的密集市场（西科尼和霍尔，1996；安德森等，2004；卡诺等，2007；斯拉姆斯基等，2005）。如果这些利益影响经济稳定，那么人们可能会想到估计人口密度系数应在我们的模型之中。教育是衡量人口年龄在25岁及以上的完成了高中学业的比例，并反映了人力资本积累水平。它类似于在拉詹和津加莱斯（1998），利文（2000年）和Cetorelli和Gambera（2001）等人先前的研究中经济增长的措施，并以平均增长研究利率理论之间的联系（特莱斯，2005）

54、。一个关于教育的措施已被用于至少一年的生长波动研究（贝卡尔特等，2006）。此外，教育被认为是与美国大都会区的部门就业的Ohuallachain和Satterthwaite（1992）增长有关。如果教育促进经济稳定良好，其回归系数的估计应该是否定的。家庭收入中位数可以初步反映经济表现的第一矩收敛作用的措施，如巴罗和萨拉-伊-马丁（1992）指出。同样的逻辑并不适用于二阶矩措施的经济表现，呈现这个变量的符号和系数的估计这一个开放性经验问题的意义。然而，类似的变量（初步的人均国内生产总值）已用于至少一年的生长波动之前的研究（贝卡尔特等，2006）。表1总结了数据。我们的样本超过2000年的美国非大

55、都市在测量19911995年的经济表现和1987年结构的措施。大都会地区排除在外，因为它们的边界一般不与个别县紧密联系，混淆从县级数据中得出的变量的测量问题。虽然没有在表中报告，两者的平均规模和人均设立机构之间的两两相关系数为-0.21和0.18，在批发行业和制造业中只有-0.07和0.07。这些小型和高度的相关性表明，结构变量的两类反映了我们样本中独立统计的产业结构方面。性能数据的选择经确定的几年后，虽然随数据结构有助于降低反向因果关系的可能性，但共同所有的实证研究增长，因果关系不能最后确定。这种滞后结构也最大限度地减少潜在的内生性偏差，作为回归量是预先确定的。三、结果表2报告人均实际收入增

56、长率标准差的回归估计，而表3报告就业增长率的SD估计。随着工业结构在三种方式上被四个部门的每一个测量，每个表报告12个回归。由于缺少或零数据，如表所显示：在每个部门设立的几个县利用各种回归观测数字略有不同。显著性水平计算被从社会企业计算的异方差修正。在表2中，各种结构的措施是统计在12个回归中的八个显著。研究发现，人均收入在制造业中更稳步增长，零售和批发单位雇用更多的工人，或者是由批发单位的年发货量测值较大。这些发现与上述讨论的营业额的假设是一致的。相反，稳定的增长被认为出现在雇用更少工人的批发单位或那些批发单位较少的城市或更多的人均制造场所。批发业的对比估计可能反映在非线性关系基础上超出了本

57、研究范围的问题。这些结果表明，产业结构是系统相关的收入增长率的波动性的方式，在很大程度上是一致的，但在一些主要行业有所不同。表2中方程的拟合是温和的，调整范围从0.3到接近0.5。多变量控制大部分在0.01显著水平上。初步住户收入中位数是从来不显著的，相对于初步人均国内生产总值在贝卡尔特等（2006）研究中被发现的一些对消费增长波动性的回归。收入增长在人口较多的县更加稳定。人口密度和教育有助于提高人均收入增长率，虽然当控制人均批发场所时，教育会失去它的意义。教育的可变性与贝卡尔特（2006）等在越野范围中发现的相对比。在表3，各种结构措施在12个统计回归中的八个再次显著。在四个部门的任何一个有

58、更多人均机构的地方，就业被看到的是更多的稳步增长，这与工作匹配假说相一致。此外，稳定的就业增长发生在那些雇用很多工人的批发或服务机构，或在一个有较大的平均美元生意额的零售或服务的场所，这与营业额的假设一致。这些结果表明，后者在大公司利益的工作寿命超过弥补理论上的波动越来越集中的决策，如上所述。这些方程拟合得比表2更紧，调整系数大约0.1。人口密度从来不显著，但其他控制变量却始终具有重要意义。在人口较多的县，就业增长更加稳定，其中教育更高，或者最初的家庭收入中位数更低。人口结果与上面讨论的就业选配的假设，以及教育和初步结果匹配的收入对消费增长的波动越野回归由贝卡尔特等（2006年）发现的是一致的

59、。如表2，当控制人均批发机构时，教育就失去了意义。四、结论在之前有关产业结构对经济稳定结果的建议和重要预测的启发下，我们也探讨了在多个部门的具体构造措施和增长率在人均收入与就业的实际波动的实证联系。在三分之二的规格处被发现显著相关，控制了一组标准的环境变量。总体而言，营业额假设被发现主宰低效集中决策的对比预测，正如就业和收入都倾向于稳步增长的较大的地方机构。这些结果表明抵消造福于编制第一时刻鉴定费用由谢弗（2002，2006a，b）鉴定。就业选配假设是符合那些人均更多机构运作的稳定增长，正如发现的在各行业的就业回归和在制造业的收入回归找到的相一致。对比结果发现，批发单位的收入回归是通过就业人数和人均批发场所来衡量。虽然我们的多年滞后结构减少了产业结构和增长之间波动的可能性反向因果关系，使得因果关系不能被最后确定，这一缺点是所有的实证增长研究所共同的。因此，我们应谨慎地制定政策推论，有待观察的进一步研究应在确切机制的模式背后进行。