Judging from the progress of endogenous growth theory, it is still in an active development period. Although there is no epoch-making innovation, many achievements have been made in the combination of modern methods and classical theories, such as the establishment of a quantitative model of the relationship between R & :D(R & development) investment and economic growth, and the re-exploration of Schumpeter's creative destruction. In addition, in the empirical analysis, although there are still many unsolved problems, some achievements have been made in estimation methods, variable adjustment, data adjustment, qualitative factor quantification and so on. From the perspective of future development, the development of endogenous growth theory will go in two directions: one is to follow the route of nonlinear dynamic model and simulate the real economic world more accurately with more complex mathematical models; The second is the research of metrological inspection.
After entering the 1990s, economists have deepened their research on endogenous growth theory and made new progress.
These advances are mainly reflected in the refinement of the original endogenous growth model. For example, since Romer (1986) put forward externalities, economists have conducted more in-depth research on the endogenous sources of economic growth. For example, Romer (1990) regards technological progress as the expansion of intermediate products, and assumes that this expansion comes from individual optimization decisions. Young( 199 1) put forward a limited model of learning by doing, in which learning by doing is limited, so growth may be restricted by invention. Young( 1993) proposed an endogenous growth model in which the number of intermediate products and final products expanded simultaneously, in view of the substitutability and complementarity of intermediate products. Oritigeira(2000) introduced leisure into the endogenous growth model driven by human capital. Due to the introduction of leisure, the utility function is no longer monotonous, which leads to the existence of multiple equilibrium and the instability of growth path. Basu and Weil( 1999) put forward a growth model, which links technology with a specific K/L ratio. In this model, growth is driven by two aspects, on the one hand, learning by doing (with a specific K/L ratio), and on the other hand, technological progress and capital accumulation (technological progress needs to change the K/L ratio). Jones( 1995, 1999), Dinopoulos et al. (1999), Young( 1998) and Segerstrom( 1998) discussed the scale effect in the economic growth model. They think that the growth models established by Romer/Aghion and Howitt/Lucas all contain scale effect, but this scale effect has no empirical support. However, if the scale effect is removed from the above models, the long-term nature of the model will be fundamentally changed, and the growth may also change from endogenous to exogenous. Therefore, they are committed to establishing a growth model without scale effect. Chol-Won Li(2000) made a new thinking on endogenous growth by establishing a research and development department composed of two departments, namely, the research and development department to improve the quality of new products and the research and development model to increase the types of intermediate products. In his model, endogenous growth (that is, growth without population growth) requires harsh conditions, and semi-endogenous growth is a more reasonable explanation. In some recent literatures, such as the special issue of Journal of Economic Theory (200 1) on endogenous growth model and nonlinear relationship, economists have begun to discuss the chaotic path or bifurcation point in the process of endogenous growth, and introduced the study of economic growth into the nonlinear direction; In addition, some economists have discussed the uncertainty in growth, especially when there are multiple equilibrium points in the growth path (such as Kuzumino (200 1)).
The revival of neo-Schumpeter doctrine
Another important development of endogenous growth theory in 1990s was the revival of Neo-Schumpeter. Since Aghion and Howitt put forward the role of creative destruction in the growth process in 1992, in their book "Endogenous Growth Theory" published in 1998, they used a lot of space to make a detailed analysis of Schumpeter's method and the creative destruction of technological progress. Aghion and Howitt introduced the concept that new technology makes the original technology obsolete in their model, thus making technological progress a creative destruction process. Another feature of Neo-Schumpeter is related to the micro-mechanism of technological progress. In the literature of growth theory in 1990s, many models [12] developed the relationship between market structure and technological progress (such as Aghon and Howitt (1998)). But as far as I can see, it is worthwhile for economists to establish an endogenous model of technological progress in the market structure [13].
The idea of promoting economic growth by division of labor, represented by Yang Xiaokai, is gradually recognized by mainstream economists because of the introduction of corner solution and the modeling of division of labor and growth. The research based on division of labor has not yet occupied the mainstream position in the study of economic growth. The relationship between division of labor and economic growth [14] was first put forward and emphasized by Smith system, and then carried forward by A.Young( 1928), but it was not until the 1980s that economists paid attention to it again. Economists study the concept of division of labor and its relationship with growth along two lines: the first line is based on the deepening of production circuitous degree and from the perspective of manufacturers' optimal decision-making, such as A.Young, Romer( 1987) and Grossman (19 1, 198). The second view is based on the fact that the division of labor is the result of the optimal choice of agents in the economy, which is reflected in the level of individual specialization. Baker (1992), Yang Xiaokai (199 1, 1992, 1993) and others studied this idea. Analysis along the first line of thought, the main problem lies in ignoring the emergence and evolution of division of labor. Although Romer (1987) takes the kind of intermediate products as the circuitous degree of production, and assumes that the intermediate products are not completely replaceable and complementary (in fact, the production function in the model used by Romer is a D-S production function), a general equilibrium solution of the circuitous degree of production is solved. However, the increase of transaction cost caused by the increase of production circuitous degree is not discussed in the analysis process, and its model is of little practical significance. Yang Xiaokai's contribution is that he realized the defect that Romer model ignored the transaction cost, linked the division of labor with the transaction cost, and explained the evolution of the division of labor with the transaction cost. But judging from his published papers and monographs (such as Yang Xiaokai (1998,2000,2001)), an important issue in Yang Xiaokai's system is the concept of evolution. In Yang's view, the division of labor is an evolutionary process, and it is the transaction cost that determines the evolution of the division of labor. However, the problems in the definition and measurement of transaction costs limit the application of Yang's model. What is more fatal is that, in Yang's view, the rationality of the subject is problematic, that is, he has unlimited rational reasoning ability at a given transaction cost, but he knows nothing about predicting the evolution of transaction costs [15].
With the development of theory, many economists have realized that the biggest problem facing endogenous growth theory is how to make empirical analysis. No matter the research route along Romer's independent R&D department or Lucas' research route on human capital spillover, they are faced with the problem of how to conduct empirical analysis.
From the research point of view, this kind of empirical research is actually carried out along two technical routes. First, conduct inter-country research to find evidence of endogenous growth; The other is to study the economic growth factors along a country's long-term data; Or discuss the effect of a specific factor, such as opening to the outside world, taxation, equality, financial progress, long cycle, education expenditure, innovation and so on. About economic growth.
Most studies along the first technical route are essentially based on the famous Barrow regression, that is, taking the per capita income growth rate of a country as the dependent variable and the per capita income of a country as the independent variable to test whether the national income growth rate converges. For example, Barro( 1995, 1996) tested the convergence trend of 92 countries, American states and Japanese counties; Kremer's (1993) research on the process of economic growth in various countries shows that there is a positive correlation between economic growth and population size, which supports the endogenous growth theory empirically. Michael J. boskin (200 1) made an empirical study on postwar economic growth. He believes that technological progress should be reflected in the adjustment of both human capital and material capital. Based on this, he came to the conclusion that the contribution rate of technological progress to GDP growth is above 50%, while the tangible capital is above 25%. The decline in growth rate after the 1970s should be attributed to the technological progress of pure material capital adjustment. Greenwood et al. (1998, NBER, W6647) calculated the economic growth of the United States after World War II, and thought that the growth of the United States had a strong correlation with technological progress. At the same time, they believe that there is a strong complementarity among human capital, technological progress and capital improvement in the process of economic growth, and they have obtained some evidence of endogenous growth. The problem faced by Aghon and Howitt (1998) is how to find data that can represent the differences between countries. For example, the difference in growth rate between one country and another may be caused by the cultural traditions and political and economic systems of the two countries [16]. At the same time, it is also worth discussing whether the GDP gap between countries is as big as the exchange rate shows. However, for GDP, more importantly, the research along the first technical route did not find much empirical evidence to support the endogenous growth theory. For example, Delong and Summers (199 1) studied the fact of American economic growth, and showed that the growth of equipment investment is an important factor of economic growth. The research of Mankiw, Romer and Weil (1992) (the famous MRW test) shows that Solow-Swan model with diminishing returns and exogenous technological progress can explain the economic growth rate, and their work also shows the existence of conditional convergence; Young( 1995) calculated the total factor productivity growth rate of newly industrialized countries in East Asia by using the production function beyond logarithmic polymerization. Young's research shows that the TFP growth rate of newly industrialized countries in East Asia, such as Hongkong, Singapore, China, Taiwan Province Province and South Korea, is very low during the period of high-speed economic growth, which is not enough to explain its growth over ordinary developing countries. Yang believes that the economic growth of these countries comes from a large amount of capital. Jones (1995) studied the role of R&: D in productivity growth. He found that after the war, OECD countries R&: The sharp increase in expenditure has no substantial impact on the improvement of its productivity. Dinopous and Thompos( 1999) investigated the scale effect in economic growth and concluded that there is no empirical fact to support the scale effect. All these indicate that endogenous growth theory is not completely consistent with empirical facts. As Temple( 1999) pointed out, the empirical research on growth largely supported the types of Solow's growth model in 1950s. Sala-I-Martin(200 1) also points out that Solow growth model can explain convergence better than AK model.
The research along the second technical route has achieved a wide range of results, although there is still a lack of consistency between these results in terms of the role of individual factors in economic growth. Abigit. V.Banerjee's Inequality and Growth (NBER Working Paper No.7793), Douglas Holtz-Eakin's Intergenerational Conflict, Human Capital Accumulation and Economic Growth (NBER Working Paper No.7762), Paul Beaudry and David Green's Population Growth, Technology Application and Economic Output (200 1), S.EDWARDS's "Openness, Productivity and Growth" (1997+0997) [Based on the data of 93 countries, this paper studies the robustness of the relationship between openness and TFP growth and draws a positive conclusion] [197] Engen et al. (1996)'s "Taxation" The relationship between tax revenue and economic growth is empirically studied by country analysis and micro-analysis methods, and it is concluded that tax revenue is almost neutral in the short term, but the cumulative result is significant in the long term. The factors affecting economic growth are analyzed in detail. But on the whole, as Ben Fine(2000) pointed out, the empirical analysis of many endogenous growth theories is still similar to the old total factor productivity (TFP) analysis method, but the input factors are explained more widely to include the factors that generate growth, so that the growth generated under the incomplete market background can be described more deeply on a micro basis [18].
Another direction of the development of new economic growth theory is to study the interactive relationship between economic growth and structural change. For example, John Laitner(2000) discussed the relationship between structural change and economic growth. He believes that in the process of industrialization, a country's savings rate rises endogenously, so the economic growth rate also changes. This view of John leitner is actually an important extension of the two-sector (or multi-sector) growth model that appeared after 1990s.
Judging from the progress of endogenous growth theory in 1990s, endogenous growth theory is still in an active development period, such as the research on R&; D the establishment of the quantitative model of the relationship between input and economic growth, and the re-exploration of Schumpeter's creative destruction. In addition, in empirical analysis, as Ben Fine(2000) pointed out, the empirical study of growth theory faces the following three problems: the independence of variables (the test of the model assumes the independence of data, but in fact the variables are interactive and independence cannot be guaranteed); The data is based on the selectivity of the model, ignoring the growth process and focusing on the results of growth; The model uses random variables to match the randomness of data and variables, but the real data is the result of various random shocks. However, in the 1990s, the adjustment of estimation methods (such as the breakthrough of Barro regression) and variables (such as the long-term data of many countries, see Summers-Heston (1988/19911995), as well as the adjustment of education data, From the perspective of future development, the development of endogenous growth theory will be carried out in two directions. The first direction is to take the route of nonlinear dynamic model and simulate the real economic world more accurately with more complex mathematical models; The other direction is the research of econometric test, including introducing more variables, adjusting variables to make them conform to reality, quantifying qualitative factors and so on.