MP build a profit-maximizing neoclassical model of optimal firm size and growth across different industries based on differences in industry fundamental and firm productivity.
贡献:
第一,Conglomerate firms are less productive than single-segment firms of a similar size, for all but the smallest firm. (与diversification discount的文献一致)
Q1: Spinoff的sub通常比较小,与MP是巧合,或者什么关联吗?
第二,The productivity pattern within a conglomerate firm's segments is consistent with our simple value-maximizing model.
MP发现大公司里的大子公司large segments比smaller segments要高效。
第三,MP examine the growth and investment of conglomerate firms across the industries in which they operate.
Q2: Conglomerate firms grow more in industries in which their plants are productive when that industry receives a positive demand shock and when their other segments receive negative shocks. (什么正的负的shock?哪里来的shocks?)
Ans: Shock是指某产品的需求突增或突减。
在一个行业里,如果一个single-segment公司非常高效,它要diversify的机会成本(opportunity cost)就非常高!所以total size一样的情况下,均衡的解应该是single-segment firm而非conglomerate.比较优势(comparative advantage)也决定了larger segment比smaller segment高效。
Model也预测了the effect of demand shocks on the growth of a conglomerate's segment depends on the segment productivity.同样的demand shock, 肯导致高效的部门扩大生产力,同时导致低效的部门卖掉剩余的生产力。所以大企业里的这两种部门,在面临同样的demand shock时,反应是不同的。
文章的一个重要假设:大企业部门面临的demand shock也会影响企业其他部门的growth rate,即使没有agency cost和financial market imperfections.如果demand shock增加了某些高效部门的growth rate,那么它会减少这个大企业其他部门的growth rate.(那些inefficient internal capital market与MP的Neoclassical model在这里的假设不同,以示区别。)
他们发现:比较大的部门,通常比较小的部门,生产力productivity更高,这符合利润最大化行为maximizing behavior,最终各部门的marginal returns一样。各部门间的productivity不同,也说明了organizational talent has an industry-specific component.但是,如果把部门在大企业中的顺序按照size排列,并且和其他大企业按次序排列的部门比较(说的通顺一点,就是A企业的a1,a2,a3,a4,a5和B企业的b1,b2,b3,b4,b5来比较),MP发现productivity和the number of segments正相关。这说明了,那些跨行业比较多的大企业,有更高的general ability,比起那些只在几个行业内运营的企业。
但是,conglomerates也可能选择suboptimal方案,来进军一个他没有specific skill的行业。为了区别这种可能性,MP examine how conglomerates grow in different industries. MP test whether the growth rate of a segment is related to the productivity of conglomerates' segments and how growth changes in response to industry shocks.
Q3: 我没懂上一段,他们的test和他们想区别的两种可能性,有什么关联???
结果显示the growth of both more- and less-productive firm segments is related to productivity and fundamental industry factors, both in recession and expansion periods.大企业的部门grow more slowly如果其他部门more productive,但是faster如果其他部门Less productive.
MP也发现一些evidence of agency cost.他们检测了那些重组(experience restructuring)的大企业。MP发现broken-up的大企业和他们的model of optimal growth不一致。(Q4: 也就是neoclassical model吗???)但是即便是这些公司,MP也没有发现大企业subsidize他们那些less productive的部门。他们发现the growth rates of firms that remain conglomerates(大多数conglomerates都这样) are strongly sensitive to industry variables and productivity.这些发现与optimal behavior一致。这些结果也表明surviving的大企业在他们运营的各行业中都很有效率(grow efficiently across industries they operate).
Lang and Stulz (1994) and Berger and Ofek (1995)两篇著名diversification discount papers. Comment and Jarrell (1995) 发现stock market returns to conglomerate firms比较低。Berger and Ofek (1995)和Comment and Jarrell (1995) 用agency theories来解释,公司内部的internal capital分配inefficient. Lang and Stulz (1994)认为poorly performing may choose to become conglomerates (self selection). Lamont (1997) 和 Shin and Stulz (1998) 检验了Tobin's q和现金流的关系。Scharfstein (1997)发现大企业的managerial ownership越低,在low Tobin's q的企业中的投资就越大。Rajan, Servaes, and Zingales (2000)发现the extent of firm investment in segments in low-q industries is related to the diversity of investment opportunities across segments (Q5: 看不懂这话!!!) 但是,这些研究都有一个缺陷。他们都用single segment的Tobin's q来替代大企业面临的投资机会(investment opportunity).他们这样就假设了single-segment firms和conglomerates面临的投资机会是一样的。
I. Optimal Firm Size and Growth
假设(assume):在每个行业内,有些公司比另一些公司更有效(more productivity).
假设:对于任何一个经理人,他管理一个小公司,都比大公司,要搞的好。
MP model里,公司在他们拥有比较优势的行业里运营。假设1(predictions):how equilibrium diversification across industries of conglomerates depends on the organizational talent and the correlation of abilities across industries. 假设2: how optimal growth in different industries is affected both by the comparative advantage of the firm in each industry, and by changes in industry conditions.
A. The First-Best Equilibrium with Conglomerate Firms
在I, III,公司最佳方案是single-segment firm,但是在区域II,公司最佳方略是diversify.
Proposition 1: Versification is optimal if θvB>vA and θvA>vB.
所以,公司是否多元化,部分取决于公司在两个行业内的比较优势。同时,这个结论也预示了,大部门比小部门更高效(more productive)。
Productivity和Focus的关系,取决于both on the distribution of ability within these firms and on the distribution of ability across firms.想象一下,如果一个CEO在某个行业内特别牛逼,他就会猛投那个部门。但如果一个CEO在行业A和行业B的能力差不多,而且这两个行业本来也差不多,那么就应该多元化。
1) Take the number of industries to be 10. 设定10个行业。
2) 假设25,000公司,它们每个都被赋予一个firm-specific ability在这10个行业内。
3) high ability就预示着high productivity
4) ability assignment d 遵照正态分布,平均值1,离散0.5
5) output, input prices和cost都是固定值,p=200, u=5, w=2
Example 1: Firm ability is industry specific. The draws are independent.
Example 2: firm-specific effects: The draws within a firm for each of the 10 industries are correlated. Common ability遵循正太分布,平均值0和离散0.5.
把EX1中的random industry ability和EX2中的common ability相加。
1) Determine每个公司的最佳运营industry, the amount of production in each industry, given the price of inputs and outputs.
2) 追踪那些只在一个行业,两个行业,等等运营的公司
3) 得到simulation
看上图,segment number,应该是segment number rank,可以理解为,在一个大企业中按照rank来排列,最前面的segment得到的重视越多。
Page 731, MP do not differentiate their model from other models, such as empire-building models, which predict that firms inefficiently expand into industries outside their core competence.
但MP用demand shock来区分他们与inefficient internal capital market model的确别。
Just make fun of Empire Building
B. Shocks and Growth of Conglomerate Firms
来分析the effects of price shocks on firm size
1) One period and two dates: t=1,2
At t=1, 公司预料到下一个时间,会有一个价格冲击∆pA,对于output level,在行业A。
2) 行业内的公司可以买或卖capacity,其价格为rA。
3) Following output price shock,所以也会有一个adjustment of price in capacity —— ∆rA。
4) 公司可以(a)用尽所有的capacity来生产,(b)卖掉一部份capacity,用剩余的生产,(c)还可以购买更多的capacity来生产。
5) 公司也会可以买卖capacity in industry B来回应demand shock in industry A.
6) 这些adjustment也会导致rb和pb的变动,即capacity 价格和output价格的变动。
7) 假设,供求弹性不是perfectly elastic,所以high level of rA会买capacity,反之,low level会卖。
8) t=2,公司实现现金流。假设没有残值no salvage value。
MP derive the demand for capacity as a function of the market prices for the output good pi, i=A,B and the prices of capacity ri.
output price is exogenous
capacity price is endogenous
1) 先来看看conglomerates中的segment production是对自己部门,或者其他部门的price shock进行反馈的。
解偏微分方程,得到dA*∆pA-∆rA。
很明显,和output price正相关,和capacity price负相关。最终取决于d(ability)的大小。如果前面的benefit大于后面的cost,就增加在行业A中的产量。如果大企业中的部门对于demand shock的反应是增加产量,那么他们的机会成本就增大了。机会成本在这里指the opportunity cost of its organizational talent.
这就使得这个大企业在行业B中的optimal productivity下降了。
2) The effect of a price shock in industry A on marginal producers is more ambiguous.价格冲击对于边际的影响,不太好说。扩张产量会提高capacity的价格。如果这个影响可以忽略不计,那么当然应该扩张。
但是,如果capacity提高的吓人,反而是卖掉一些capacity更赚钱,那么marginal producers in industry A最佳选择是卖,然后投产到行业B中去。
Proposition 2 (Demand Shocks and Focus): A positive price shock in industry A provides incentives for conglomerates that are productive in industry A to focus in that industry. If the supply of capacity in industry A is sufficiently inelastic, conglomerates that were marginal producers in industry A have an incentive to focus in industry B. (就是output price提高,产量应该增大。但是如果投产的要素价格很刚性,就是说,)
Corollary 1: 一个大企业在一个行业内的产量越高,这个行业的price shock对它的影响越大。
Proposition 3 (Demand Shocks and Focus): Give a positive price shock in industry A, a conglomerate that is more productive in industry A will grow more slowly in industry B than an otherwise similar conglomerate that is less productive in industry A. Given a negative price shock in industry A, the relative growth rates of the two conglomerates in industry B are reversed.
(如果 行业A有一个price shock,大企业#1在行业A中有很大的比重,大企业#2在行业A中less productive,那么#1在行业B中的扩张(grow)要比#2慢很多。如果这个price shock是负的,那么#1在行业B中会快,而#2慢一些。)
C. Relation of Our Model to the Agency Literature
MP考虑了三种情形:
第一,the agents may invest optimally, but divert a portion of the proceeds for their own benefits as higher overhead at the firm level or as over-payments for acquisitions.经理人通体投资是最佳的,但是有一部分钱,他们花在了更高的overhead花费,或者并购时买贵了。MP的方法没有办法测出这种情形,因为在这个情形下,大致的公司增长还是符合利润最大化的。
第二,the agents may have a private benefit from investment in capacity.这种情况下,公司会over invest, 但最高效的部门(segment with the highest marginal return)仍然获得很多资源。如果所有的公司都over invest,也很难说他们跟comparative statics of the model with respect to demand shocks and differences in technology有什么区别。
第三,opportunistic behavior by agents may cause firms to misallocate resources across segments. 这种misallocation有可能因为(1)公司内部矛盾Internal conflicts,(2)公司扩张去了他们不熟悉的行业。
支持第三种观点的文献有:
Q2: Conglomerate firms grow more in industries in which their plants are productive when that industry receives a positive demand shock and when their other segments receive negative shocks. (什么正的负的shock?哪里来的shocks?)
Ans: Shock是指某产品的需求突增或突减。
在一个行业里,如果一个single-segment公司非常高效,它要diversify的机会成本(opportunity cost)就非常高!所以total size一样的情况下,均衡的解应该是single-segment firm而非conglomerate.比较优势(comparative advantage)也决定了larger segment比smaller segment高效。
Model也预测了the effect of demand shocks on the growth of a conglomerate's segment depends on the segment productivity.同样的demand shock, 肯导致高效的部门扩大生产力,同时导致低效的部门卖掉剩余的生产力。所以大企业里的这两种部门,在面临同样的demand shock时,反应是不同的。
文章的一个重要假设:大企业部门面临的demand shock也会影响企业其他部门的growth rate,即使没有agency cost和financial market imperfections.如果demand shock增加了某些高效部门的growth rate,那么它会减少这个大企业其他部门的growth rate.(那些inefficient internal capital market与MP的Neoclassical model在这里的假设不同,以示区别。)
他们发现:比较大的部门,通常比较小的部门,生产力productivity更高,这符合利润最大化行为maximizing behavior,最终各部门的marginal returns一样。各部门间的productivity不同,也说明了organizational talent has an industry-specific component.但是,如果把部门在大企业中的顺序按照size排列,并且和其他大企业按次序排列的部门比较(说的通顺一点,就是A企业的a1,a2,a3,a4,a5和B企业的b1,b2,b3,b4,b5来比较),MP发现productivity和the number of segments正相关。这说明了,那些跨行业比较多的大企业,有更高的general ability,比起那些只在几个行业内运营的企业。
但是,conglomerates也可能选择suboptimal方案,来进军一个他没有specific skill的行业。为了区别这种可能性,MP examine how conglomerates grow in different industries. MP test whether the growth rate of a segment is related to the productivity of conglomerates' segments and how growth changes in response to industry shocks.
Q3: 我没懂上一段,他们的test和他们想区别的两种可能性,有什么关联???
结果显示the growth of both more- and less-productive firm segments is related to productivity and fundamental industry factors, both in recession and expansion periods.大企业的部门grow more slowly如果其他部门more productive,但是faster如果其他部门Less productive.
MP也发现一些evidence of agency cost.他们检测了那些重组(experience restructuring)的大企业。MP发现broken-up的大企业和他们的model of optimal growth不一致。(Q4: 也就是neoclassical model吗???)但是即便是这些公司,MP也没有发现大企业subsidize他们那些less productive的部门。他们发现the growth rates of firms that remain conglomerates(大多数conglomerates都这样) are strongly sensitive to industry variables and productivity.这些发现与optimal behavior一致。这些结果也表明surviving的大企业在他们运营的各行业中都很有效率(grow efficiently across industries they operate).
Lang and Stulz (1994) and Berger and Ofek (1995)两篇著名diversification discount papers. Comment and Jarrell (1995) 发现stock market returns to conglomerate firms比较低。Berger and Ofek (1995)和Comment and Jarrell (1995) 用agency theories来解释,公司内部的internal capital分配inefficient. Lang and Stulz (1994)认为poorly performing may choose to become conglomerates (self selection). Lamont (1997) 和 Shin and Stulz (1998) 检验了Tobin's q和现金流的关系。Scharfstein (1997)发现大企业的managerial ownership越低,在low Tobin's q的企业中的投资就越大。Rajan, Servaes, and Zingales (2000)发现the extent of firm investment in segments in low-q industries is related to the diversity of investment opportunities across segments (Q5: 看不懂这话!!!) 但是,这些研究都有一个缺陷。他们都用single segment的Tobin's q来替代大企业面临的投资机会(investment opportunity).他们这样就假设了single-segment firms和conglomerates面临的投资机会是一样的。
I. Optimal Firm Size and Growth
假设(assume):在每个行业内,有些公司比另一些公司更有效(more productivity).
假设:对于任何一个经理人,他管理一个小公司,都比大公司,要搞的好。
MP model里,公司在他们拥有比较优势的行业里运营。假设1(predictions):how equilibrium diversification across industries of conglomerates depends on the organizational talent and the correlation of abilities across industries. 假设2: how optimal growth in different industries is affected both by the comparative advantage of the firm in each industry, and by changes in industry conditions.
A. The First-Best Equilibrium with Conglomerate Firms
Proposition 1: Versification is optimal if θvB>vA and θvA>vB.
所以,公司是否多元化,部分取决于公司在两个行业内的比较优势。同时,这个结论也预示了,大部门比小部门更高效(more productive)。
Productivity和Focus的关系,取决于both on the distribution of ability within these firms and on the distribution of ability across firms.想象一下,如果一个CEO在某个行业内特别牛逼,他就会猛投那个部门。但如果一个CEO在行业A和行业B的能力差不多,而且这两个行业本来也差不多,那么就应该多元化。
1) Take the number of industries to be 10. 设定10个行业。
2) 假设25,000公司,它们每个都被赋予一个firm-specific ability在这10个行业内。
3) high ability就预示着high productivity
4) ability assignment d 遵照正态分布,平均值1,离散0.5
5) output, input prices和cost都是固定值,p=200, u=5, w=2
Example 1: Firm ability is industry specific. The draws are independent.
Example 2: firm-specific effects: The draws within a firm for each of the 10 industries are correlated. Common ability遵循正太分布,平均值0和离散0.5.
把EX1中的random industry ability和EX2中的common ability相加。
1) Determine每个公司的最佳运营industry, the amount of production in each industry, given the price of inputs and outputs.
2) 追踪那些只在一个行业,两个行业,等等运营的公司
3) 得到simulation
看上图,segment number,应该是segment number rank,可以理解为,在一个大企业中按照rank来排列,最前面的segment得到的重视越多。
Page 731, MP do not differentiate their model from other models, such as empire-building models, which predict that firms inefficiently expand into industries outside their core competence.
但MP用demand shock来区分他们与inefficient internal capital market model的确别。
Just make fun of Empire Building
B. Shocks and Growth of Conglomerate Firms
来分析the effects of price shocks on firm size
1) One period and two dates: t=1,2
At t=1, 公司预料到下一个时间,会有一个价格冲击∆pA,对于output level,在行业A。
2) 行业内的公司可以买或卖capacity,其价格为rA。
3) Following output price shock,所以也会有一个adjustment of price in capacity —— ∆rA。
4) 公司可以(a)用尽所有的capacity来生产,(b)卖掉一部份capacity,用剩余的生产,(c)还可以购买更多的capacity来生产。
5) 公司也会可以买卖capacity in industry B来回应demand shock in industry A.
6) 这些adjustment也会导致rb和pb的变动,即capacity 价格和output价格的变动。
7) 假设,供求弹性不是perfectly elastic,所以high level of rA会买capacity,反之,low level会卖。
8) t=2,公司实现现金流。假设没有残值no salvage value。
MP derive the demand for capacity as a function of the market prices for the output good pi, i=A,B and the prices of capacity ri.
output price is exogenous
capacity price is endogenous
1) 先来看看conglomerates中的segment production是对自己部门,或者其他部门的price shock进行反馈的。
解偏微分方程,得到dA*∆pA-∆rA。
很明显,和output price正相关,和capacity price负相关。最终取决于d(ability)的大小。如果前面的benefit大于后面的cost,就增加在行业A中的产量。如果大企业中的部门对于demand shock的反应是增加产量,那么他们的机会成本就增大了。机会成本在这里指the opportunity cost of its organizational talent.
这就使得这个大企业在行业B中的optimal productivity下降了。
2) The effect of a price shock in industry A on marginal producers is more ambiguous.价格冲击对于边际的影响,不太好说。扩张产量会提高capacity的价格。如果这个影响可以忽略不计,那么当然应该扩张。
但是,如果capacity提高的吓人,反而是卖掉一些capacity更赚钱,那么marginal producers in industry A最佳选择是卖,然后投产到行业B中去。
Proposition 2 (Demand Shocks and Focus): A positive price shock in industry A provides incentives for conglomerates that are productive in industry A to focus in that industry. If the supply of capacity in industry A is sufficiently inelastic, conglomerates that were marginal producers in industry A have an incentive to focus in industry B. (就是output price提高,产量应该增大。但是如果投产的要素价格很刚性,就是说,)
Corollary 1: 一个大企业在一个行业内的产量越高,这个行业的price shock对它的影响越大。
Proposition 3 (Demand Shocks and Focus): Give a positive price shock in industry A, a conglomerate that is more productive in industry A will grow more slowly in industry B than an otherwise similar conglomerate that is less productive in industry A. Given a negative price shock in industry A, the relative growth rates of the two conglomerates in industry B are reversed.
(如果 行业A有一个price shock,大企业#1在行业A中有很大的比重,大企业#2在行业A中less productive,那么#1在行业B中的扩张(grow)要比#2慢很多。如果这个price shock是负的,那么#1在行业B中会快,而#2慢一些。)
C. Relation of Our Model to the Agency Literature
MP考虑了三种情形:
第一,the agents may invest optimally, but divert a portion of the proceeds for their own benefits as higher overhead at the firm level or as over-payments for acquisitions.经理人通体投资是最佳的,但是有一部分钱,他们花在了更高的overhead花费,或者并购时买贵了。MP的方法没有办法测出这种情形,因为在这个情形下,大致的公司增长还是符合利润最大化的。
第二,the agents may have a private benefit from investment in capacity.这种情况下,公司会over invest, 但最高效的部门(segment with the highest marginal return)仍然获得很多资源。如果所有的公司都over invest,也很难说他们跟comparative statics of the model with respect to demand shocks and differences in technology有什么区别。
第三,opportunistic behavior by agents may cause firms to misallocate resources across segments. 这种misallocation有可能因为(1)公司内部矛盾Internal conflicts,(2)公司扩张去了他们不熟悉的行业。
支持第三种观点的文献有:
- Lamont (1997),石油行业产生positive price shock时,那些大企业的其他部门(非oil industry)在被补贴。
- Shin and Stulz (1998) find that investment of conglomerates' segments is affected by cash flows in other segments.
- Scharfstein and Stein (2000) imply that weaker segments get subsidized by stronger ones.差的部门被好的补贴。
- Rajan et al. (2000) argue that the investment by conglomerates with diverse opportunities depends on internal conflicts, so that they do not predict that such conglomerates will concentrate their growth in their relatively most productive segments.
V. Conclusions