资本价格与经济结构外文文献翻译中英文2020

资本价格与经济结构外文文献翻译中英文2020
英文
The relative price of capital and economic structure
Roberto Samaniego
Abstract
Are trends in the price of capital technological in nature? First, we
find that trends in the relative price of capital vary significantly across
countries. We then show that a multi-industry growth model, calibrated to
match differences in economic structure around the world and
productivity growth rates across industries, accounts for this variation –
mainly due to variation in the composition of capital. The finding
indicates that the rate of change in the relative price of capital can be
interpreted as investment-specific technical change – the extent to which
productivity growth is relatively more rapid in the capital- producing
sector. The model also accounts for the empirical dispersion of
investment rates, but not of rates of economic growth.
Keywords: Investment-specific technical change, Multi-sector
growth models, Structural transformation, Capital goods prices
Introduction
Declines in the relative price of capital are viewed as an important
factor of economic growth in the United States (US). See for example
work by Hulten (1992), Greenwood et al. (1997), Cummins and Violante
1

(2002) and Oulton (2007). These studies typically identify the decline in
the price of capital as being technological in nature, reflecting faster
productivity growth in the production of new capital than in the
production of consumption and services – a phenomenon known
as investment-specific technical change (ISTC). However, the extent to
which the relative price of capital declines in other countries is not known.
In addition, it is not known whether trends in the price of capital around
the world can be given a technological interpretation, such as ISTC. An
alternative hypothesis is that these differences are due to the presence of
barriers to capital accumulation, as proposed by Restuccia and Urrutia
(2001) to account for differences in levels of the price of capital.
We begin by documenting that the rate at which the relative price of
capital changes over time varies significantly across countries. We find
that the median growth rate of the price of capital is zero. In addition, the
price of capital increasesin as many places as it decreases. This indicates
that, if there is a technological explanation for this phenomenon, technical
progress in capital relative to other sectors must vary widely around the
world.
If the explanation is indeed technological, however, one would
expect such glaring differences in productivity to be evidence of
draconian barriers to international technology transfer (or trade). The
alternative possibility is that capital and consumption are themselves
2

highly disaggregated, and that there are substantial differences in
the composition of capital and consumption around the world that
account for the aggregate differences in the trends in the relative price of
capital.
We ask whether this variation can be accounted for by differences
in industry composition. The reason we do this is as follows. It is well
known that rates of technical progress in the US differ significantly not
just between capital and non-capital, but also across types of consumption,
services and capital. Thus, even if productivity growth rates are constant
across countries for each industry, the rate of change in the relative price
of capital may be different if the composition of capital – or the
composition of consumption and services – is different. Indeed, we find
that the composition of capital is skewed towards high-TFP growth
capital types in countries where the price of capital declines rapidly. We
therefore ask: to what extent can differences around the world in industry
composition account for variation in the rate at which the relative price of
capital changes?
To this end, we employ a canonical multi-industry growth model. In
the model, the composition of the economy evolves as a result of changes
in prices of different goods or services that agents consume, as well as
changes in the prices of different capital goods. In turn, these are
determined by differences in productivity growth rates across industries.
3

We calibrate the model using detailed productivity growth data from the
US, as well as data on the initial composition of economies around the
world in the year 1991. We use constant productivity growth rates for a
given industry in all countries partly because of data limitations; however,
as mentioned, significant barriers to technological transfer would have to
exist to significantly deviate from this assumption. Composition is a key
part of the “no barriers” hypothesis.
Strikingly, we find that the model delivers a close match to the rate
of change in the relative price of capital, as measured using the Penn
World Tables (PWT) version 7.1. In a statistical sense, the model can
account for the entirety of the magnitude of variation of the growth rate in
the relative price of capital over the period from 1983 to 2011, simply
based on industry TFP growth rate differences and on differences in
industry composition across countries. Not only does the model match the
extent of variation, but also the correlations between model-generated
capital price growth rates and those in the data are highly significant. We
conclude that differences in the relative price of capital around the world
can be interpreted as a technological phenomenon – ISTC – and that a
key factor behind these differences is industry composition.
The link between composition and the decline in the relative price of
capital could be for two reasons: differences in the composition of capital,
or in the composition of non-capital. We refer to these possibilities as
4

the capital hypothesisand the consumption hypothesis, respectively. We
study the importance of each hypothesis by removing productivity growth
differences in the capital producing industries, and then separately
removing them in the non-capital producing industries. We find that
the capital hypothesis is mainly responsible for cross-country variation in
ISTC: removing productivity growth in non- capital makes very little
difference to the results, whereas removing productivity growth in
capital-producing industries results in model- generated statistics that bear
little relationship with the data.
Finally, we ask to what extent a growth model driven solely by these
factors can account for differences in aggregate behavior across countries
over the sample period. Specifically, we look at investment rates and rates
of economic growth. This is a non-trivial task, as it requires solving for
investment patterns in a model where conditions for a balanced growth
path do not hold in general. We find that the model generates investment
rates that are strongly correlated with investment rates in the PWT 7.1
data and the PWT 9.1 data, although they underpredict the extent of
empirical variation in investment rates. Thus, the model is able to capture
cross-country variation in both ISTC and (to a lesser extent) investment
rates, solely based on differences in industry composition. However, the
model does not generate a good match to variation in rates of economic
growth in the PWT 7.1 data, nor in the PWT 9.1 data. We conclude that
5

there is widespread divergence in the rate of ISTC around the world, and
that this accounts for variation in investment, but that economic growth
rates are due to other factors. Interestingly, when we give each country an
aggregate productivity trend that exactly matches its economic growth
rate in the data, investment rates are no longer correlated with those in the
data, suggesting that whatever factors do underlie rates of economic
growth are not simply captured by a trend in productivity.
The results contribute to a long-standing debate regarding whether or
not changes in the efficiency of investment are an important factor of
growth. This debate goes back to Solow (1962), Abramovitz
(1962) and Denison (1964). Greenwood et al. (1997) find that, in the US,
more than half of economic growth can be accounted for by ISTC in a
general equilibrium growth accounting framework. We provide a clear
answer to the question about whether differences in the relative price of
capital can be attributed to barriers or to technological factors, indicating
that changes in the efficiency of investment are an important factor
affecting growth rates. This is not to say that there is no scope for barriers
to be important for the relative price of capital; however, their impact
might not be direct, but rather indirect, through their influence on
economic composition. More broadly, this suggests that future work on
the manner in which factors of economic growth might be affected by
policy through the channel of economic composition could be fruitful.
6

Discussion
Comments on institutions
We find that the model without barriers accounts well for the
empirical magnitude and variation in log?gq, solely on the basis of
economic composition. On the other hand, our findings do leave the door
open for institutional factors or other barriers to
influence log?gq indirectly, through any impact they might have on
economic composition.
What might these determinants be? There is a precedent in the
literature for the idea that policy or institutional factors may affect
composition. For example, Samaniego (2006) shows in an open-economy
context that labor market regulationcan affect comparative advantage in
industries depending on their rate of ISTC, skewing industrial
composition towards industries that use capital types with low values
of gi (an effect termed high-tech aversion). Also, Ilyina and Samaniego
(2012) show that when technology adoption requires external
financing, financial underdevelopment also skews industrial composition
towards low- tech industries. This begs the question as to whether any
policy or institutional indicators might be statistically related to our
findings. Of course, there is a question of reverse causality: political
economy considerations imply that countries that depend on
technological transfer rather than de novo innovation for growth might
7

adopt particular kinds of institutions, see for example Boldrin and Levine
(2004). Given this, we briefly explore whether there is suggestive
evidence of a link between log?gq in the data and institutions, without
taking a stand on the direction of causality.
Following Samaniego (2006) we look at firing costs (drawn from the
World Bank, firing costs paid by workers with at least one year's
tenure, FC). We also look at other forms of regulation that have been
found to be important for aggregate outcomes – namely product market
regulation, measured using entry costs paid as a share of GDP, EC, as
reported by the World Bank. See Moscoso-Boedo and Mukoyama (2012).
Another possibility suggested by Ilyina and Samaniego (2012)is financial
development, which we measure using FD, the credit-to-GDP ratio, as
in King and Levine (1993). Data on FC, EC and FD are from the World
Bank 1960–2010.
In addition, Acemoglu and Johnson (2005) and others argue that
financial development is ultimately derived from the state of contracting
institutions and property rights institutions. We measure the strength of
contracting institutions using the negative of the index of legal system
formality from Djankov et al. (2003), which we call CONT. We measure
property rights enforcement using the index developed by the Property
Rights Alliance (2008), PROP, averaged over the available period
2007–2013. Finally, we also look at intellectual property rights, which
8

have been related to the generation and diffusion of technology,
see Samaniego (2013) for a survey. We measure intellectual property
rights IPR, using the patent enforcement method developed in Ginarte
and Park (1997), as reported by the World Bank, averaging over the
available sample. Ilyina and Samaniego (2011) find that copyright
enforcement specifically is a form of IPR enforcement that bears the
strongest relationship to financial development – see also Samaniego
(2013). The BSA (Software Alliance) publishes the rate at which
unlicensed software is used in different countries. Following the Property
Rights Alliance (2008), we take this measure (times ?1) as an indicator of
copyright enforcement. Finally, we also look at human capital, HC, using
the standard Barro and Lee (2013) schooling- based measure averaged
over the period. While this is not an institutional measure as such it is an
important country characteristic which could be related to the need or
ability to produce or import high-tech capital goods.
Comments on trade
The model abstracts from international trade. Eaton and Kortum
(2001) find that machinery is often imported by developing countries,
which might suggest that the price of capital could be significantly
affected by trade rather than by domestic output, and that domestic output
shares might not be indicative of the composition of capital. However
their data is for 1985, so it is not clear that their findings are relevant for
9

the relative price of capital in more recent data. Indeed, using data for
1995, Caselli and Wilson (2004) find that the composition of imported
machinery is very highly correlated with the composition of
domestically-produced capital, both in developed and developing
countries. Nonetheless, it would certainly be interesting to explore the
extent to which trends in the price of capital might be affected either by
trade flows or by changes in trade costs. In particular, Mutreja et al.
(2018) argue that reductions in trade costs may lower the relative price of
capital by allowing countries to more easily access capital from countries
that might produce them more efficiently. This suggests that one factor
that might contribute to the dispersion in investment rates could be trade
costs.
Concluding remarks
We document extensive differences in the rate of change in the
relative price of capital around the world. We then show that these
differences can be accounted for on the basis of differences around the
world in economic composition, without recourse to any barriers or
frictions. We also find that a general equilibrium model economy
accounts for a significant portion of the variation in the rate of change in
the relative price of capital and for differences in investment rates around
the world, although not for differences in rates of economic growth. We
conclude that these differences can be given a technological interpretation,
10

based on differences in composition among industries with different rates
of technical progress. As a result, the term “investment-specific technical
progress,” which the literature widely identifies with declines in the
relative price of capital, is appropriate. Given the key role played by
industry composition in this phenomenon it seems important to
understand what are the deep determinants of industry composition. Is it
due to comparative advantage or other trade- theoretic mechanisms? Is
due to policy distortions, as suggested by Samaniego (2006) or Ilyina and
Samaniego (2012)? Or does it result form hysteresis, for example, based
on the date at which the process of development began in earnest and the
speed of transition, as in Ngai (2004)? These are likely useful questions
for further research.

中文
资本相对价格与经济结构
罗伯托·萨曼涅戈
摘要
资本价格的趋势本质上 是技术性的吗？首先，我们发现各国的资
本相对价格趋势差异很大。然后，我们表明，为适应世界各地经 济结
构的差异和各行业生产率的增长而进行了校准的多行业增长模型，可
以解释这种变化-主要 是由于资本构成的变化。该发现表明，资本相
对价格的变化率可以解释为特定于投资的技术变化，即在资 本生产部
11

门中生产率增长相对较快的程度。该模型还考虑了投 资率的经验分
散，但没有考虑经济增长率。
关键字：特定于投资的技术变革，多部门增长模型，结构转型，
资本价格
引言 资本相对价格的下降被视为美国经济增长的重要因素。参见格林
（Greenwood）等人的Hu lten（1992）的著作。 （1997），康明斯和
维奥兰特（2002）和奥尔顿（2007） 。这些研究通常将资本价格的下
降识别为技术性的，这反映出新资本生产中生产力的增长快于消费和服务业生产中的增长-这种现象被称为投资特定技术变革（ISTC）。但
是，尚不清楚其他国家的 资本相对价格下降的程度。此外，尚不清楚
是否可以对国际资本价格趋势进行技术解释，例如ISTC。 另一种假
设是，这些差异是由于存在资本积累壁垒，如Restuccia和Urrutia
（ 2001）提出的，用以解释资本价格水平的差异。
我们首先记录下来，各国的资本相对价格随时间变 化的速率差异
很大。我们发现，资本价格的中位数增长率为零。另外，资本的价格
在减少的地方 增加。这表明，如果对此现象有技术上的解释，相对于
其他部门，资本在技术上的进步在世界范围内必须 相差很大。
但是，如果这种解释确实是技术性的，人们会期望生产率的这种
明显差异将成为国 际技术转让（或贸易）的严峻障碍的证据。另一种
可能性是，资本和消费本身被高度分解，世界各地的资 本和消费构成
存在实质性差异，这说明了资本相对价格趋势的总体差异。
12
< /p>

我们问这种变化是否可以由行业构成的差异来解释。我们这样做
的原因如下。众所 周知，美国的技术进步速度不仅在资本和非资本之
间存在显着差异，而且在消费，服务和资本的类型之间 也存在显着差
异。因此，即使各国对于每个行业的生产率增长率是恒定的，但如果
资本的构成（ 或消费和服务的构成）不同，则相对资本价格的变化率
也可能不同。确实，我们发现，在资本价格迅速下 降的国家中，资本
的构成偏向于高全要素生产率增长的资本类型。因此，我们问：在世
界范围内 ，行业构成的差异可以在多大程度上解释资本相对价格变化
的速率变化？
为此，我们采用了规 范的多行业增长模型。在该模型中，经济构
成是由代理商所消费的不同商品或服务的价格变化以及不同资 本产
品的价格变化而演变的。反过来，这些是由各行业生产率增长率的差
异决定的。我们使用来 自美国的详细生产率增长数据以及1991年全
球经济的初始构成数据来校准模型。我们在所有国家地区 使用给定
行业的恒定生产率增长率，部分原因是数据限制；但是，如上所述，
要大大偏离这一假 设，就必须存在技术转让的重大障碍。组成是“无
障碍”假设的关键部分。
令人惊讶的是，我们发现该模型与使用Penn World Tables（PWT）
版本7 .1测得的资本相对价格的变化率非常匹配。从统计意义上讲，
该模型可以简单地基于行业TFP增长率 差异和整个行业的行业构成
差异来解释1983年至2011年期间资本相对价格增长率的整体变化幅< br>度。国家。模型不仅与变化程度匹配，而且模型生成的资本价格增长
13