有什么心得分享一下？

占位，持续更新。。

经济增长的经验研究是一个争议很大的领域。采取什么模型、选择哪些变量、采用什么方法，经济学家争议纷纷。于是产生了一个专门研究这些问题的领域——增长计量。

核心议题
选择什么模型、哪些变量、什么方法。
1. 增长计量着力在采取什么模型、选择什么变量、采用什么方法三个方面实现发展。为此，发展了贝叶斯和伪贝叶斯方法，处理模型的不确定性；使用交互项、非线性、半参方法，使得参数具有异质性。但这些方法要求大数据。
2. 增长计量从关注经济因素，到政治、社会和文化要素对经济增长的作用。

挑选内容

增长计量文献集中于两个领域：趋同研究、经济增长影响因素研究。

第一部分：趋同研究

关于经济趋同，有结构变量与初始条件谁重要之争。这取决于初始条件前的估计系数beta是正还是负。而正确的估计依赖于模型的选择、变量的选取和内生性的处理，文章4.2节着力处理这些问题。

1 内生性问题会导致估计系数的符号改变。

Endogeneityraises a second identification issue with respect to the relationship betweenbeta-convergence and economic convergence. Focusing on the Solow regressors,the value of beta can fail to illustrate how initial conditions affect expectedfuture income differences if the population and saving rates are themselvesfunctions of income. Hence, beta>=0 may be compatible with at least partialeconomic convergence, if the physical and human capital saving rates depend onthe level of income. In contrast, beta<0 may be compatible with economicdivergence if the physical and human capital accumulation rates for rich andpoor are diverging across time. Cohen (1996) argues that the conventional humancapital accumulation equation, in which accumulation is proportional to percapita output, is misspecified, failing to account for feedbacks from the stockof human capital to the accumulation process. (资本的积累不仅与产出有关，而且与自身的规模有关)。 Cohen (1996) concludes a properaccounting for the dependence of human capital accumulation on initial capitalstocks reconciles condition beta-convergence with unconditional beta-divergencefor a broad cross-section.

趋同分析的一点启发：用研究国家趋同的方法研究城市趋同问题。

2  4.3.3distribution dynamics (这部分内容不熟悉)

（1） One way of implementing the methodof distribution dynamics is to assume that the process describing the evolutionof the distribution is time-invariant and first-order Markov. Quah (1993, 1996)takes this method.

Quah D. Empiricalcross-section dynamics in economic growth. European Economic Review, 1993,37(2-3), 426-434.

Quah D. Empiricsfor economic growth and convergence. European Economic Review, 1996, 40(6),1353-1375.

（2） An alternative formulation ofdistribution dynamics that avoid discretization problems is proposed by Quah(1996, 1997) and models the cross-country income distribution at time t withdensity function.

Quah D.Convergence empirics across economies with some capital mobility. Journal ofEconomic Growth, 1996, 1(1), 95-124.

Quah D. Empiricsfor growth and distribution: stratification, polarization, and convergenceclubs. Journal of Economic Growth, 1997, 2(1), 27-59.

3 转折点的存在使得趋同的研究变得复杂和困难。

Perron (1989) thefailure to allow for structural breaks when testing for unit roots can lead tospurious evidence in support of the null hypothesis that a unit root ispresent. The role of breaks in time series convergence tests is systematicallystudied in Li and Papell (1999). Allowing for trend breaks reduces the numberof country pairs that fail to exhibit convergence. Carlino and Mills (1993)study US regions and reject convergence except under specification that allowfor a trend break in 1946. The presence of the regime break is presumablysuggestive of an absence of convergence.

第二部分：经济增长决定因素

增长计量的研究集中在经济增长决定因素的研究上。主要有以下三个方面：

A given variabledoes or does not help explain cross-country growth differences;

Heterogeneity ingrowth

Nonlinearities inthe growth process

5.1 specifyingexplanatory variables in growth regressions

影响经济增长的因素非常多，具体可见第1篇、第7篇文献，但选择什么变量呢？

One approach toresolving the problem of model uncertainty is based on identifying variableswhose empirical importance is robust across different model specifications.

四种方法：

Method 1: extremebounds analysis．The policymaker must have minimumpreferences with respect to model uncertainty.

Method 2: averagingthe statistical significance level. A variable whose sign and statisticalsignificance holds across 95% of the different models estimated is regarded asrobust.

Method 3:general-specific modeling methodologies.

Method 4: modelaveraging techniques. The basic idea is to treat the “true” growth model as anunobservable variable. In order to account for this variable, each model in themodel space is associated with a posterior model probability.

Brock, Durlaufand West (2003) 指出模型的四种不确定性。理论的不确定、变量说明的不确定、测度的不确定、异质性。

5.3 Nonlinearity and multiple regime (需要进一步看文献，大矿)

分类方法

(1) An earlycontribution to this literature is Durlauf and Johnson (1995) who useclassification and regression tree (CART) methods to search for nonlinearitiesin the growth process. The CART procedure identifies subgroups of countriesthat obey a common linear growth model based on the Solow variables. The numberof subgroups and the boundaries for the variable intervals that define them arechosen by an algorithm that trades off model complexity and goodness of fit.  

Durlauf, S.,Johnson, P. (1995). “Multiple regimes and cross country growth behaviour”.Journal of Applied Econometrics 10 (4), 365–384.

(2) Hanson (1999)threshold effect model

(3) Tan (2004)employs a procedure known as GUIDE (generalized, unbiased interaction detectionand estimation) to identify subgroups of countries which obey a common growthmodel.

Tan C. No onetrue path: uncovering the interplay betweengeography, institutions, andfractionalization in economicdevelopment. Journal of Applied Econometrics, 2010, 25(7), 1100-1127.

(4) projectionpursuit methods. Desdoigts (1999) uses these methods to separate the roles ofmicroeconomic heterogeneity and initial condition in the growth experiences ofa group of countries and identifies groups of countries with relativelyhomogenous growth experiences based on data about the characteristics andinitial conditions of each country. The idea of projection pursuit is to findthe orthogonal projection of the data into low-dimensional spaces that bestdisplay some interesting feature of the data. A well-known special case ofprojection pursuit is principal components analysis.

(5) Bloom,Canning and Sevilla (2003) based on the observation that if long-run outcomesare determined by fundamental forces alone, the relationship between exogenousvariables and income levels ought to be unique. If initial conditions play arole there will be multiple relationships.

Bloom,D., Canning, D., Sevilla, J. (2003). “Geographyand poverty traps”. Journalof Economic Growth 8, 355–378.

(6) Canova (2004)introduces a procedure for panel data that estimates the number of groups andassignment of countries or regions to these groups, drawing on Bayesian idea.

Canova, F.(2004). “Testing for convergence clubs in income per capita: Apredictive density approach”. International Economic Review 45 (1), 49–77.

6 Econometric issues I: Alternative data structures

6.1 time series approaches

这一节着力分析了时间序列分析中存在三大问题：

数据可获得性

变量的变化幅度

经济周期和波动对变量的影响。

6.2 panel data

讨论了FE的两大核心缺陷：没有考虑“个体间变化”的信息，致使估计系数的标准误增大；容易放大测度误差的作用。

解决办法：

（1）dummy

（2）the most widely-used alternativestrategy is to difference the model to eliminate the fixed effects, and thenuse two stage least squares or GMM to address the correlation between thedifferenced lagged dependent variable and induced MA(1) error term.

但是GMM也有缺陷。第一，GMM估计可能有偏；第二，GMM估计的一个假设是误差项不相干；第三，需要考虑参数异质性。

Phillips and Sul(2003) allow for heterogeneity in parameters not only across countries, butalso over time.

Phillips, P.,Sul, D. (2003). “The elusive empirical shadow of growthconvergence”. Cowles Foundation Discussion Paper No. 1398.

6.3 Event study approaches

总结：事件研究的两个纬度

（1） 前后的差异

（2） 控制组和处理组的差异

参数估计的两大问题：反向因果和参数异质。

6.4 endogeneity and instrument variables

An obvious andfrequent criticism of growth regressions is that they do little to establishdirection of causation. Explanatory variables are endogenously determined.

怎么办？

（1） One approach is to model as manyas possible of the variables that are endogenously determined. Establish structuralequations. This approach has some important advantages in both economic andstatistical terms. From a purely statistical perspective, if the structuralequations are estimated by jointly by methods such as three stage least squaresor full information maximum likelihood, this is likely to bring efficiencygains.

（2） IV.

Whether are theseinstruments valid? When the instruments are invalid, estimates will of coursebe inconsistent.

Discussions ofthe validity of instruments suffer from some degree of imprecision because ofthe need to make qualitative and subjective judgments.

7 Econometric issues II: Data and error properties

7.1 Outliers

这个问题虽小，但是数据处理的第一步。贴出两篇文献

Temple, J.(1998). “Robustness tests of the augmented Solow model”. Journal of AppliedEconometrics 13(4), 361–375.

Temple, J.(2000b). “Growth regressions and what the textbooks don’t tell you”.Bulletin of Economic Research 52 (3), 181–205

7.2 measurementerror

When there aremultiple explanatory variables, but only one is measured with error, thentypically all the parameter estimates will be biased.

7.5 cross-sectionerror correlation

也谈了空间计量问题

Brock, W.,Durlauf, S. (2001b). “Interactions-based models”. In:Heckman, J., Leamer, E. (Eds.), Handbook of Econometrics, vol. 5.North-Holland, Amsterdam

楼主，你每天读一篇吗？还是以前读过的，重新总结的？

正在读啊，每天一篇完不成。
你如果喜欢读书，可以发到论坛上文章名。我们一起讨论。
欢迎交流

最后部分：The future of growth econometrics

Pritchett (2000) lists three questions for growth researchers to address:
(1) what are the conditions that initiate an acceleration of growth or the conditions that set off sustained decline?
(2) what happens to growth when policies – trade, macroeconomic, investment – or politics change dramatically in episodes of reform?
(3) why have some countries absorbed and overcome shocks with little impact on growth, while others seem to have been overwhelmed by adverse shocks?

In all three cases, it is clear that econometric work should be informed by detailed studies of individual countries. Too much empirical growth research proceeds without enough attention to the historical and institutional context.
Another reason for advocating case studies is that much of the empirical growth literature essentially points only to reduced-form partial correlations. There is plenty of scope for informative work that tries to isolate mechanisms by which variables such as financial depth, inequality, and political institutions shape the growth process.
A more extreme view is that growth econometrics should be supplanted by the calibration of theoretical models.

() The need for a tighter connection between theory and evidence is especially apparent in certain areas. The workhorse model for many empirical growth papers continues to be Solow-Swan, a closed economy model which leaves out aspects of interdependence that are surely important. Howitt (2000) has shown that growth regression evidence can be usefully reinterpreted in the light of a multi-country theoretical model with a role for technology diffusion. More generally, there is a need for researchers to develop empirical growth frameworks that acknowledge openness to flows of goods, capital and knowledge. Here especially, research that draws on the quantitative implications of specific models, as in the work of Eaton and Kortum (1999, 2001) on technology diffusion and the role of imported capital goods, appears to be an important advance.

() The neglect of open economy aspects of the countries under study is mirrored elsewhere. Much of the empirical literature uses a theoretical framework that was originally developed to explain the growth experiences of the USA and other developed nations. In developing countries, these could include the potentially important roles of agricultural employment, dualism, and structural change, and in some cases, extensive state involvement in production.

Distinguish between different types of growth and their distributional consequences. Agenor (2004) considers some of the relevant issues, and again this appears to be a vital direction for future research.

() Ideally, research along these various lines will utilize not only statistics, but also the power of case studies in generating hypothesis, and in deepening our understanding of the economic, social and political forces at work in determining growth outcomes. Case studies may be especially valuable in two areas. The first of these is the study of technology transfer. Why are some countries more successful than others in climbing the ladder of product quality and technological complexity? What are the relative contributions of human capital, foreign direct investment and trade? The second area is the study of political economy. The two-way interaction between economics and politics.

Optimism or pessimism?
One reason for optimism is the potential that developed model averaging methods. Model uncertainty.

Another reason is the quality of data is to improve over time.
The construction of proxies is likely to make increasing use of latent variable methods. These aim to reduce a set of observed variables to a smaller number of indicators that are seen as driving the majority of the variation in the original data. Factor analysis or extracting principal components from various dimensions of variables.

The evidence found in qualitative sources such as historical narratives and studies by country experts.