Lost Decades in Growth Performance

Causes and Case Studies

Y. Chu

• Publisher: Palgrave Macmillan
• Year: 2015
• Print ISBN:
  9781349502332
  9781137478740
• eBook ISBN:
  9781137478757
• Topic: Economics
• Subtopic: Finance
• Index: Economics

There have been many noticeable incidents of 'lost decades' in economic growth, occurring in countries across the world. It has been found that in many economies, the lost decade phenomenon persists, even after the conventional set of contributing factors such as per capita income, fertility rate, life expectancy, rule of law, educational attainment, ratio of investment to national income, and openness have been taken into account. This book explains where and how these lost decades in economic growth occur in the world. The authors identify that dominant macroeconomic factors contributing to their occurrence are an abnormal supply of credits relative to national income, and poor demand management. The study pays special attention to the cases of Japan, South Korea and Taiwan, exploring their specific cases and analyzing contributing factors. While Japan suffered from excessive credit prior to the bubble bursting, and from insufficient domestic demand subsequently, Korea's growth has been stunted through structural imbalances between and within industries, as well as through changes in the orientation of public policies from growth to equality. Adversely, reduced economic growth in Taiwan has led from its populism-ridden democracy and mass media, as well as from internal disputes over national identity. Lost Decades in Growth Performance provides a revealing insight into the factors affecting economic growth across the world, and will be an invaluable resource for anyone with an interest in global and Asian economics. It also offers a fundamental source of reference for students and academics in general equilibrium models, economic development and East Asian economies.

1

Excessive Credits and the “Lost Decades” in Growth Performance

Yun-Peng Chu

1.1 Introduction

The term “lost decade” probably comes from the title of one of the short stories of the famous “lost generation” novelist, F. Scott Fitzgerald, written in 1939. The story is about a man who lost most of his memories from the last decade, but did not know why. In the end we find out that he had been drunk for ten years. The term became popular as a description of the Japanese experience from the late 1980s to the 1990s, when growth stagnated at low levels.¹ Recently, the term has also been applied to other economies, including the United States.²

This chapter will use the 1969–2010 data from 28 large economies to first identify possible candidates for those suffering from a lost decade, defined as a period of length of about ten years with shrinking or stagnating growth of per capita GDP, occurring in the past 20 or so years. It will then explore the reasons for this shrinkage or stagnation by econometrically estimating a set of cross-country panel growth equations.

The equations contain the conventional set of variables, namely: the initial level of per capita GDP, education, fertility, expected length of life, ratio of investment to GDP, rule of law, and openness (see e.g., Barro and Sala-i-Martin, 2003). It also contains variables not conventionally covered, including oil shocks, financial crises and their lagging values, and the deviation from the “norm” of the ratio of domestic credit to nominal GDP.

The main findings of this study are: (i) most of the conventional set variables have the expected signs and are significant; (ii) oil shocks are detrimental to the growth of heavy oil importers who had to cut back on consumption as a result; (iii) financial crises will lower the growth rate but not permanently; (iv) the deviation of credit growth from its norm will have negative effects on growth. Finding (iv) is consistent with the “quantity theory of credit” (Werner, 2005). This theory emphasizes that the traditional quantity theory of money should be replaced by the quantity theory of credit as the credit level is the variable which truly affects the allocation of resources in the economy. The ratio of the level of credit to nominal GDP should be stable. Whenever the ratio is kept at much higher (or lower) levels than the norm for a prolonged period of time, it creates bubbles (or stagnation) and in either case growth will be slower as a result. The findings of this chapter support Werner’s quantity theory of credit.

For financial crises, the result does not fully support the, so-called, “balance sheet recession” parable (see, e.g., Koo, 2011). A balance sheet recession is said to occur when, after the bursting of a major financial or real estate bubble, people try to repay their debt to normalize their balance sheet no matter how low the interest rate is. During such a recession, interest rate policies are ineffective as individuals have become irrationally conservative with respect to borrowing. In other words, people change their behavior from profit maximizing to winding down the debt, at whatever the cost and for a long, long time. Koo’s view is not supported by the results of this 28-country study since the effects of financial crises on growth are of limited duration, and they have no permanent effects on growth.

Section 1.2 briefly reviews the existing literature on growth equations and explains the basic structure of the model, including definitions of the variables. Section 1.3 reports on the main results and Section 1.4 interprets them in the light of the quantity theory of credit and the theory of balance sheet recession. The concluding remarks and the related plan for the work in this book are presented in Section 1.5.

1.2 The growth equation: expanded to include economic crises and credit growth

Econometric research on cross-country growth performance has been a popular endeavor among scholars and has undergone a series of changes in its orientation. Most works before the mid-1990s investigated the effects of various economic policies,³ followed by those focusing on the influence of adverse geographic conditions,⁴ and finally by those advocating the “rule of institutions.”⁵ In this large-scale search for the determinants of economic growth, 145 different kinds of variables, belonging to 37 broad categories according to a survey by Durlauf, Johnson and Temple (2005), have appeared on the right-hand side of growth equations.

To explain the cross-country and cross-period variations in the growth of per capita GDP, the following basic equation is used:

where g_pcgdp is the growth rate of real (and PPP-based; see Heston, Summers and Aten, 2009) GDP per capita, EDU is a measure of educational attainment, LIFE_EXPECT is the inverse of life expectancy at age one, FERT is the fertility rate, RULAW is a measure of the rule of law, INVGDP is the ratio of investment to GDP, RCTARIFF is a measure of tariffs on imports, FINANCIAL_CRISIS and OIL_SHOCK are dummies for financial and oil crises respectively, and finally PERIOD_n is a period dummy, which takes the value of 1 for the samples in period n (n = 1975–1979, 1980–1984, 1985–1989, 1990–1994, 1995–1999, 2000–2004 and 2005–2008).⁶ The definition of the ratio of credit to GDP (CREDIT_GDP) has been given above and need not be repeated here.

The model in Equation (1.1) differs from other cross-country studies in several important aspects. Firstly, only countries with a population of more than ten million in 1960 and for which data exist for 1969–2010 are included in the sample. This limits the observations to 28 economies, constituting about 70 per cent of the world’s population.⁷ Such a loss in the number of cross-section samples is a trade-off resulting from the inclusion of specific variables, such as the occurrence of financial crises, the data for which are hard to compile.⁸

Secondly, in the regression, the traditional approach that arbitrarily chooses a cutoff point and divides the entire period for which data is available into several fixed intervals, usually decades or five-year periods, is not adopted here. To better identify the long-term growth rate trends, a five-year moving average of per capita GDP growth is chosen instead. This allows enough data points to be created to not only mitigate the downsides of having fewer sample countries in the regression, but also to allow for country-specific variances of the disturbances and their possible correlation across countries.⁹ By the same token, the five-year moving average of all the explanatory variables and their instruments are taken.¹⁰ The data cover the years from 1969 to 2010, with the middle year of the moving averages ranging from 1971 to 2008.

The tariff regime is represented by the “tariff rate class” data from the Economic Freedom of the World. This ranges from 0 to 10 (converted to 0 to 1 in the model) with 10 representing a totally free trade regime and 0 indicating an effective tariff rate of at least 50 per cent.¹¹ In the study’s estimation, this variable is found to capture the effects of openness better than the traditional ratio of imports plus exports to GDP (filtered for the effects of geographic size and population).¹²

As for shocks, it is clear from numerous studies¹³ that the shock of large-scale rising oil prices can suddenly swell import bills, drain foreign exchange reserves, unleash inflationary pressure on the economy and curtail the purchasing power of ordinary citizens, which eventually result in a lower growth rate. It is quite possible that shocks on the scale of the first and second oil crises in the mid-1970s and early 1980s would have an effect on the growth rate that could not be averaged out over five years. Therefore, oil price shock dummies are included in the equation.

It is known that during the two oil crises there were two peaks in oil prices: one in 1974 and another in 1980. It is also known that for most economies, the downside effects occurred about one year following these peaks and the worst of these effects lasted for about two years. So in this study, an oil shock dummy of 1 is assigned to oil importing countries that suffered from a fall in oil consumption (−1 for all others) in 1975–1976 and 1981–1982 for the two crises. When taking the five-year moving average, the dummy value equals the total number of occurrences during that five-year period.¹⁴

Financial crises have also been identified as a major culprit behind faltering economies.¹⁵ These also often result in a downward shift in a country’s growth rate that cannot be averaged out over five years. Specifically, a country is labeled 1 if it is enmeshed in any one of the three types of financial crisis, banking, currency or sovereign debt (as defined by Laeven and Valencia, 2008), in any given year. When taking the five-year moving average, the same principle of the accumulated number of occurrences applies.

Finally, Werner (2005) argues that the financial variable that really impacts on the allocation of resources and can be measured unequivocally is the level of credit available to businesses. It should be kept in line with the movement of nominal GDP. If its level relative to nominal GDP is much higher (or lower) than the norm, a financial bubble (or crunch) is likely to occur, and the economy is likely to experience bubble-bursting (stagnation). In either case, growth will be slower.

In this chapter the norm is defined as the band of one-standard-error deviation from the long-term linear trend observed between total domestic credit and nominal GDP. Abnormal fluctuation is therefore defined as the value of the actual ratio of domestic credit to nominal GDP above or below the norm, and is measured in terms of the number of standard deviations above (or below) the band.¹⁶

1.3 The result: balanced growth of domestic credit and nominal GDP matters

Model one includes all the variables usually included in conventional studies. The results (Table 1.1) show that the logarithm of the initial level of per capita GDP has a negative effect, inferring that, other things being equal, countries starting at a lower level of economic development are expected to grow faster. It has an expected sign and is highly significant.

The parameter of the education variable is positive and significant, suggesting a higher education level enhances the growth rate, other things being equal. Similarly, life expectancy, which is defined as the inverse of life expectancy at one year of age, has a negative effect on growth and fertility rate; it inhibits growth as it preve...

Table of contents

• Cover
• Title
• 1  Excessive Credits and the Lost Decades in Growth Performance
• 2  A Third Decade of Low Growth? Lessons from Japan on Financial Management and Economic Growth
• 3  Its Not Structural Change, but Domestic Demand: Japans Productivity Growth
• 4  Long Waves of Prosperity and Decline: What Makes Financial-Led Capitalism Different?
• 5  The Korean Economy: Structural Changes and Challenges
• 6  Taiwans Lost Decades: Populism and Internal Contradiction
• 7  Paradigmatic Shifts in Policies, Evolution of the Political Economy, and the Lost Decade in Growth Performance
• Index