Category Archives: Factor Accumulation and Growth



P. Aghion, P. Howitt & D Mayer-Foulkes

Quarterly Journal of Economics, February 2005

Principal Research Question and Key Result Can differences in financial development explain why we have not observed the convergence of growth rates predicted by the neoclassical growth model? The results of the empirical analysis indicate that differences in initial income creates divergence, but that conditional upon differences in income, financial development counteracts the effect and leads to convergence.
Theory A large group of countries have converged to similar growth paths whilst the poorest have continued to diverge. Productivity differences appear to be the key factor underlying the divergence. Whilst productivity is affected by a number of factors (geography/institutions), long-lasting differences in technological process is most likely the most significant factor.  Financial constraints it is argued are the cause of the differences in technological process.Technology is created exogenously at the technological frontier and countries behind that frontier can adopt those technologies. This technology transfer requires the receiving country to invest in R&D type activity to adapt the technology to the local environment. As the frontier moves further ahead the cost of this R&D increases. However, due to agency problems receiving countries may tend to underinvest and hence fall further and further behind the frontier. This is because the investment an innovator needs in order to adopt foreign technology must be supplied by the credit markets. This innovator can defraud her creditors by hiding the results of a successful innovation at a cost that is related to the level of financial development (as better financial development means better creditor protection and enforcement). The result of this is that credit is less available generally and specifically it is available only in proportion to the innovators known income. As this income is dependent on domestic productivity, technological laggards face a disadvantage of backwardness. As less is invested in technological adoption in the presence of low financial development, this prevents convergence as low technology lowers productivity which in turn determines incomes and growth.


Motivation If poorly designed credit markets are a key reason for low growth, then policies to strengthen credit enforcement, and encourage R&D investment could spur growth. There should be an “advantage” to technological backwardness in that the further behind the technological frontier the easier it is for countries to progress simply by implementing technologies that have been developed elsewhere. This should mean that the gap is stable but not growing. This is not what we observe and the paper seeks to find out why.
Data Cross section of 71 countries over the period 1960-1995. Financial development is measured as ratio of private credit to GDP excluding credit made available by central and development banks.
Strategy The strategy compares growth in country i to the technological frontier country which is assumed to be the US. The specification is as follows:Growthi – Growth1 = B0 + BfFi + By(Yi-Y1) + BfyFi(Yi-Y1) + BxXi + Ei

i denotes country, 1 denotes frontier country (USA) F is financial development, Y is initial per capita income in 1960, X is a vector of controls.

A positive By indicates that as the difference in initial income increases [i.e. that (Yi – Y1) becomes more negative] the Gi-G1 measure will also become more negative i.e. will exhibit divergence. A negative Bfy means that conditional initial levels of income increasing financial development increasing financial development will increase [i.e. make less negative] the Gi-G1 measure.

If all countries grow at same rate irrespective of financial development then all coefficients would be 0. If all countries  converge irrespective of financial development then By would be negative as larger/small gaps in initial income mean larger/smaller differences in growth. If countries with higher financial development are more likely to converge then Bfy will be negative as per above.

They instrument for financial development using legal origins in order to get around endogeneity (richer countries can better afford to enforce and monitor credit contracts so high historic growth rates could be causing financial development rather than the other way around) and omitted variable problems.


Results The results are consistent with the theory that better financial development conditional on initial income levels will tend to create convergence i.e. convergence depends upon the level of financial development. Bfy is negative and in the full specification is equal to -0.63, and is statistically significant at the 1% level.Countries can converge only if the level of financial development F is sufficient such that it is greater than or equal to  – By/Bfy which according to the results should be around 25%
  • They control for other characteristics that are thought to determine growth such as education, inflation, size of government, political stability etc. and the results are robust.
  • They use alternate measures of financial development such as liquid liabilities and ratio of credits by banks. The results are robust.
  • They remove outliers.
  • They use alternate instruments such as settler mortality as suggested by AJR. In all cases Bfy is negative.
  • They use productivity growth as the dependent variable to confirm that financial development is affect growth through productivity rather than just capital accumulation.
  • Use of panel data would have allowed for exploitation of within as well as between country variation. This would have allowed for controls for time and country fixed effects.  However, the authors argue that levels of financial development are largely persistent and any variation is due to measurement error which would bias the coefficients to zero.
  • By excluding development banks from the measure of financial development they may be excluding a major source of financing in developing countries
  • The IV strategy is not entirely convincing.  They do not report the first stage. Legal origins could be correlated with regulatory environment, geography, human capital, all of which can drive growth. They do perform a Sargan test, but this can generally only tell us if instruments are bad, it cannot confirm that they are good, as there could be so many different endogeneity problems that they cancel each other out.
Implications It appears that financial development can be and important long run driver of growth. However, as Durlaf, Johnson and Temple point out, over 145 regressors have found to have statistically significant effects on growth. Thus the most we can say is that how important financial development is will have to be determined on a case by case basis. The estimated coefficients are averages over the whole sample and as such cannot account for the heterogeneous effects financial development on growth. Even if the results are accepted it is not clear that we know how to develop a financial system, and even if we did, it is not clear that any programme would be successful when it interacts with all the other idiosyncratic aspects of a national economy.