آیا بعد از تصویب تکنولوژی جدید رشد بهره وری سقوط کرده است؟
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|11296||2001||23 صفحه PDF||سفارش دهید||8756 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Monetary Economics, Volume 48, Issue 1, August 2001, Pages 173–195
A number of theoretical models of technology adoption have been proposed that imply that measured productivity growth may initially fall and then later rise after the adoption of a new technology. This paper investigates whether or not this implication is a feature of plant-level data from the Colombian manufacturing sector. We focus on technology adoption embodied in new equipment. We find evidence that the effect of a large equipment purchase is initially to reduce plant-level total factor productivity growth.
A number of theoretical models of technology adoption have been proposed with the following feature. After a production unit adopts a new technology, not all the expertise in the old technology transfers to the new technology and there is a period of technology-specific learning. One implication of these theories is that measured productivity growth may at first fall and then later rise after adopting a superior technology.1 In this paper we provide micro-evidence on the question of whether productivity growth first falls and later rises after the adoption of new technology embodied in new equipment. We are motivated to address this question as micro-evidence on productivity growth is key for issues related to aggregate productivity growth dynamics. Consider an example. Greenwood (1996), Hornstein and Krusell (1996) and Greenwood and Yorukoglu (1997) hypothesize that an increase in the pace of embodied technological change is the cause of the aggregate productivity growth slowdown experienced by the majority of the advanced economies since the 1970s. A microeconomic mechanism behind this hypothesis is that existing production units experience a temporary fall in productivity growth after adopting new technology embodied in new equipment.2 At the aggregate level, productivity growth could temporarily slow down when an increased fraction of production units make such investments. To evaluate such a hypothesis at a quantitative level, one would need micro-evidence on productivity growth dynamics after a production unit adopts new technology. To address the question posed above, we identify the adoption of a new technology at a particular production unit with the purchase of equipment. In particular, we will say that those plants making equipment purchases that increase their real equipment stocks by more than a critical fraction are adopting new technology embodied in new equipment. A number of remarks are in order in regards to this assumption. First, an equipment purchase is precisely the mechanism of technology adoption emphasized in the literature. Second, the evidence in the papers by De Long and Summers 1991 and De Long and Summers 1993 and Greenwood et al. (1997) suggests that equipment investment may be a quantitatively important source of technology adoption.3 Third, in plant-level data it is the case that investment displays a lumpy pattern at the plant level with the bulk of plants making little or no purchases of equipment in a given year but large percentage changes in the stock of equipment in other years.4 Thus, our measure of technology adoption is consistent with the notion that technology adoption embodied in equipment occurs somewhat infrequently at the plant level. Lastly, we realize that our measure of technology adoption is far from perfect. Our reaction to this is two-fold. First, even if this measure is imperfect we will still be addressing an interesting question (i.e. Does productivity growth fall after a large equipment purchase?). Second, we regard it as a key issue for future research to focus on data sets that potentially allow one to distinguish between equipment purchases reflecting technology adoption versus those reflecting the acquisition of more capital of a technology known to the plant. Our empirical strategy is straightforward. We focus on a data set of plants from the Colombian manufacturing sector. For each plant, we calculate total factor productivity (TFP) growth rates across time periods. We then regress the productivity growth of a plant on the current and past values of our plant-level measure of technology adoption, controlling for industry and/or time effects. We find evidence that productivity growth falls when a plant makes a large equipment purchase. More precisely, we find evidence that the effect of a large increase in a plant's stock of equipment is to reduce productivity growth by 3–9 percent in annual data. The fall in productivity growth is 3 percent when the criteria for a large equipment investment is 25 percent of the equipment stock, whereas it is 9 percent when the criteria is 100 percent of the equipment stock. We find no evidence to support the proposition that a plant's productivity level eventually rises so as to surpass the productivity level existing before the large equipment investment, after correcting for industry effects. This paper is organized in four sections. Section 2 presents a number of background facts about the data set, plant-level productivity growth and plant-level investment in equipment. Section 3 presents our main results and then discusses interpretations of these results from the perspective of a few different models. Section 4 concludes. 1.1. Related literature To the best of our knowledge, there is only scattered micro-evidence bearing on the question that we address. We discuss work in three separate areas. First, there is a literature consisting of case studies by Baloff 1966 and Baloff 1970, Russell (1968) and others in which plants from the US manufacturing sector have changed products or the production process and in which the level of productivity initially falls and then later rises. These studies are part of a large literature on learning curves. The focus of this literature has been to document the upside of the learning curve rather than any potential downside after a switch in technology. This is true, for example, of the well-known work of Bahk and Gort (1993) who estimate learning at new plants in the US manufacturing sector.5 Second, there is a literature on adjustment costs and on firm-level investment. Parts of these literatures have been surveyed by Chirinko (1993) and Dixit and Pindyck (1994). While some of the models presented in these literatures are potentially consistent with the findings of this paper, the empirical component of these literatures has not answered the question that we pose. With this said, we are aware of some work that is suggestive that firm investment in physical capital may be associated with falls in productivity. In particular, Pakes and Griliches (1984) regress firm-level, accounting profits on past investments. They find that accounting profit increases more strongly to investment lagged several periods than to more recent investment. Third, there is a literature on information technology and productivity growth, reviewed in part by Brynjolfsson and Hitt (1998) and Yorukoglu (1998). Much of this literature focuses on the issue of whether there is an information technology productivity paradox. To the best of our knowledge, existing studies have not focused on characterizing falls in productivity growth at the time of an investment in information technology. However, with this said, some of this literature has found evidence for learning by doing in the years after an investment in information technology.
نتیجه گیری انگلیسی
The main findings of the paper are as follows: (1) Our best estimate is that the contemporaneous effect of a large equipment investment is to decrease a plant's total factor productivity growth by 3–9 percent in annual data. This decrease is larger for larger critical values for what constitutes a large equipment investment. (2) This finding holds within industries and is robust to plausible amounts of unmeasured quality improvements in equipment. (3) The bulk of investment in either equipment or all reproducible physical capital occurs at existing plants rather than at new plants. We attach the following significance to these findings. First, the findings are inconsistent with models where all plants within an industry are affected by a common disembodied technology shock and/or by technological improvements embodied in new equipment. Second, if large equipment investments coincide with the adoption of new technology embodied in new equipment, then the findings imply that the adoption of new technology contemporaneously reduces total factor productivity growth. Third, we believe that the finding that the bulk of equipment investment occurs at existing plants has implications for future work. This finding suggests that the literature which attempts to quantify the implications of microeconomic models of equipment investment for aggregate productivity growth issues should concentrate on models where equipment investment occurs not only at new plants but also at existing plants. We mention two directions to explore in future work. First, what is the effect on productivity growth of equipment purchases that reflect technology adoption versus those that merely reflect the acquisition of more capital of a technology known to the plant? This is a key question. To address this question a data set that allows such a distinction to be measured is needed. Doms et al. (1997) describe a data set that is potentially promising in this regard. Second, it would be useful to see if the findings presented here are confirmed in data sets for other countries and time periods.