Does Rapid Economic Growth Accelerate Democratization? TimeSeries Evidence from High Performing Asian Economies
By
Sam Hak Kan TANG ^{a}
and
Linda Chor Wing YUNG ^{b}
ABSTRACT
We examine the direction of causality between growth and democratization for the high performing Asian economies using a new timeseries technique called autoregressive distributive lag. We find that for all eight of such economies, the direction of causality runs consistently from democratization to growth and not the other way around. Rapid growth in the high performing economies appears to have little effect on democratization. We also find that the net effect of democratization on growth is not always positive. Against the widelyheld view that growth enhances democratization, our evidence suggests that rapidly developing countries under authoritarian rule are unlikely to improve their democratic institutions.
Keywords: Growth, Democratization, High Performing Asian Economies.
JEL classification: O10, P16, O57
^{a. }Corresponding author. Economics Program, the University of Western Australia, 35 Stirling Highway, Crawley WA 6009, Australia. Tel: +61864882931. Fax: +61864881016. Email: sam.tang@uwa.edu.au
^{b. }Department of Economics, the Chinese University of Hong Kong, Shatin, NT, Hong Kong. We gratefully acknowledge the Chinese University of Hong Kong for financial support through Direct Grant 2020784.
Introduction
When the Chinese Premier, Wen Jiabao, attended the ASEAN meeting in Laos in November 2004, he was quizzed by the media about the issue of democracy and human rights in China. He commented that China has achieved a rapid rate of growth for the past twentyfive years and inevitably it would become more democratic with increasing economic prosperity. His comment echoes a norm among political leaders in high performing Asian economies (HPAEs), who argue that economic prosperity is a far more important goal than democracy, and that democracy will eventually flow from sustained economic prosperity. Their argument raises two immediate questions: whether rising income levels can accelerate the pace of democratization, and whether democratization necessarily harms growth. This paper will attempt to shed some light on the first question, which is about the determinants of democracy, and will provide clues to answering the second.
The question of whether rising income levels accelerate democracy has received little attention in the empirical literature. Some empirical studies find a strong correlation between growth and democracy (see, for example, Barro, 1996, 1999; Huntington, 1991; and Lipset, 1959). Other studies distinguish between outcomebased and demandbased measures of democracy. Minier (2001), who uses democratic movements to measure the demand for democracy, finds that such demand does not increase monotonically with the level of income, but decreases after a country reaches a certain income threshold. Recently, however, Rigobon and Rodrik (2004) find that the effect of rising incomes on democratic institutions, an outcomebased measure, is positive, but statistically insignificant.
Given the inconclusive literature, the goal of this paper is to reexamine the empirical relationship between growth and democracy using data from HPAEs. Specifically, it empirically tests whether there is a significant longrun causal relationship between growth and democracy using a relatively new timeseries approach called the autoregressive distributive lag (ARDL) bounds test.^{1} The results of this test will shed light on whether rising income levels have a significant positive effect on democracy, and whether democratization necessarily harms growth.
This paper contributes to the literature by offering empirical results that are based on individual country studies of HPAEs. First, we can check the compatibility of our countryspecific results with generalizations of crosscountry or panel data analyses. Second, a key contribution of this paper is its adoption of the timeseries approach rather than the more common crosssection or panel data approaches.^{2} Employing the timeseries approach allows the researcher to explore how changes in one variable over time affect another variable, and thus to address the issue of longrun causality between variables. In particular, use of the relative new ARDL bounds test technique eliminates the required assumption of stationarity for the underlying series in conventional cointegration studies.^{3}
Our results consistently show a significant longrun causal relationship running from democracy to growth for all HPAE. However, we fail to find that growth has a significant effect on democracy for these fastest growing economies of the world. Thus, our results do not support the widelyheld view that rising income levels accelerate democratization. We argue that both authoritative governments and the unwillingness of policymakers to yield to increasing popular demand for democracy in HPAEs account for the absence of a significant linkage between growth and democratization. In addition, consistent with the results of previous empirical studies, we find that the net effect of democratization on growth can be negative, which may result from a more egalitarian redistribution system.
The remainder of this paper is organized into five sections. In the second section, we discuss theoretical arguments for the links between growth and democracy. The third and fourth sections discuss the data and the ARDL bounds test technique respectively. The fifth section presents the empirical results, followed by a discussion of the results in the sixth section. The seventh section summarizes and concludes the paper.


Background
In the empirical literature on democracy, the two main foci are either the effects of democracy on growth or the determinants of democracy. This paper concerns both of these areas by attempting to establish the direction of causal relationship between growth and democracy. As there theoretical arguments support both directions of causation, the issue requires empirical examination.
In the theoretical literature, political stability, the quality of governance, government size, human capital, income equality, trade openness, and physical capital accumulation are believed to be the main channels through which democracy affects growth.^{4} The net effect of democracy on growth ultimately depends on the relative strength of each of these factors. Lee (2003) models income redistribution, democracy, and growth, and reports that a democratic regime grows more rapidly than an oligarchic regime if the redistribution scheme is elitist or the income distribution is less skewed. If the society is paternalistic or income distribution is highly skewed, then a democratic regime performs worse than an oligarchic regime.
Rodrik (2000) emphasizes the importance of “local knowledge” that allows the market to perform adequately. He argues that participatory political systems are the most effective for processing and aggregating local knowledge. According to him, democracy is a metainstitution for building good institutions, and there is strong evidence to indicate that participatory democracies enable higher quality growth.
Empirically, Tavares and Wacziarg (2001) find that democracy fosters growth by improving the accumulation of human capital and by lowering income inequality, but hinders growth by reducing the rate of physical accumulation and by raising the ratio of government consumption to GDP. They find that the net effect of democracy on growth is moderately negative. Borner et al. (1995) report that out of sixteen empirical studies, three show a positive relationship between democracy and growth, three show a negative relationship, and the remaining ten are inconclusive. Helliwell (1994) and Barro (1996) find a nonsignificant negative effect of democracy on growth once several growthdetermining variables are held constant.
On the effects of growth on democracy, Lipset (1959, 1960) and Huntington (1991) are the two most frequently cited sources who have argued that rising income levels cause better democratic institutions and not the other way around. Lipset points out that education plays an important role in facilitating the desire for democracy as income rises. He argues that “the higher one’s education, the more likely one is to believe in democratic values and support democratic practices” (1959, p. 79). Lipset’s contention is that rising income gives rise to improvement in human capital, which in turn serves as a catalyst for democratic institutions. Huntington (1991) also broadly defines economic development as including an increasing volume of international trade and the development of sizeable middle class, all of which facilitate the desire for increasing participation in the political process. He argues that economic integration exposes societies to democratic ideas that prevail in other parts of the world.
Empirical studies such as those of Alvarez et al. (2000), Barro (1999), Glaeser et al. (2004) and Minier (2001) support the view that growth comes before improvement in democratic or political institutions. For example, Minier (2001) finds that the demand for democracy increases with the level of income up to a certain income threshold, after which it decreases. Glaeser et al. (2004) find evidence to support Lipset’s contention that democracy is not necessary for growth. They conclude that “countries that emerge from poverty accumulate human and physical capital under dictatorships, and then, once they become richer, are increasingly likely to improve their institutions.” (2004, p. 298). Clearly, they argue for a unidirectional causality running from growth to democratic institutions.

Data
The spectacular pace of economic growth that has been achieved by HPAEs in recent decades provides a natural environment in which to test the relationship between growth and democracy. We choose eight Asian economies that have experienced a rapid rate of growth in the past three decades: Hong Kong, Indonesia, Malaysia, the Philippines, South Korea, Singapore, Taiwan, and Thailand. In Table 1, we show the real per annum growth rates for these economies. Clearly, the Philippines is an underperformer since it has an average real growth rate of 3.6 percent over the 19701996 period, which is relatively low compared to those of the four Asian tigers (Hong Kong, South Korea, Singapore, and Taiwan) and the newly industrializing economies (Indonesia, Malaysia, and Thailand). We nevertheless include it in our study because it has experienced substantial changes in democratic institutions in the last two decades. Moreover, China is another economy that deserves special attention and should be a prime candidate for study, but, unfortunately, its quarterly GDP data are highly unreliable. We thus have no choice but to exclude China from our sample.
Table 1: Average Real Per Annum Growth Rates
Country

19701996
Average

1997

1998

1999

2000

2003

Hong Kong

7.4

5.0

5.1

2.9

10.5

3.2

Indonesia

7.3

4.7

13.0

0.3

4.8

4.1

Malaysia

7.5

7.3

7.4

5.8

8.3

5.3

Philippines

3.6

5.2

0.8

3.2

4.0

4.7

South Korea

8.1

5.0

6.7

10.7

8.8

3.1

Singapore

8.6

8.2

0.4

5.4

9.9

1.1

Taiwan

8.5

6.7

4.6

5.4

5.9

3.2

Thailand

7.7

1.7

10.2

4.2

4.3

6.7

Sources: Data are from World Bank tables, GDP in constant prices, with the following exceptions. (a) Data for 2003 are from Data Stream (b) Data for Taiwan are from the National Statistics Database at http://www.stat.gov.tw/main.htm. (c) Data for Indonesia were taken from http://www.bi.go.id/bank_indonesia_english/main/statistics/. (d) Data taken from the World Bank website http://www.worldbank.org/data/countrydata/countrydata.html.
Democracy is measured by an index called “democratic accountability” that is prepared by the International Country Risk Guide (ICRG). Democratic accountability measures “how responsive [a] government is to its people, on the basis that the less responsive it is, the more likely it is that the government will fall, peacefully in a democratic society, but possible violently in a nondemocratic one.” The highest score of 5 indicates “free and fair elections for the legislature and executive as determined by constitution, viable opposition and independent judiciary.” The lowest score of 0 indicates autarchy, as defined by “leadership of the state by a group or single person, without being subject to any franchise, either through military might or inherited right.”^{5}
We adopt the measure of democracy from the ICRG for three reasons. First, it has been used extensively in the literature. Second, the ICRG has distinct categories of ‘political risk’ (such as corruption, rule of law, government stability, democratic accountability and bureaucratic quality), which allows researchers to be reasonably specific in what they are measuring. Third, it has broad coverage across countries. However, it also has limitations. First, it is only available for a relatively short time span from 1984. Second, it is prepared by analysts of the respective countries and designed mainly for international investors. Thus, it is possible that such a risk rating may be distorted in reflecting only the concerns of international investors and not of the country as a whole.
Table 2: Summary Statistics for Democratic Accountability and GDP Growth

Sample period

Democratic accountability

GDP growth (%)

Average

Std. deviation

Average

Std. deviation

Hong Kong

1984Q2 2003Q4

2.29

0.89

4.94

4.81

Indonesia

1994Q2
2003Q4

3.14

1.13

3.10

6.46

Malaysia

1992Q2
2003Q4

3.16

1.01

6.00

5.46

Philippines

1984Q2
2003Q4

4.52

0.92

2.63

4.20

S. Korea

1984Q2
2003Q4

4.06

1.77

6.81

4.33

Singapore

1984Q2
2003Q4

2.79

0.49

6.36

5.00

Taiwan

1984Q2
2003Q4

4.27

1.06

6.44

3.31

Thailand

1994Q2
2003Q4

4.06

0.61

3.41

5.94

Note: Democratic accountability (DA) is measured by an index. It has a scale from 0 to 5, with higher scores indicating the greater responsiveness of a government to its people. A score of 0 indicates autarchy and a score of 5 indicates free and fair elections. DA data are available on a monthly basis from the International Country Risk Guide. We average the monthly DA series to get quarterly DA series. GDP growth refers to the yearonyear quarterly growth rate of real GDP. Quarterly GDP data for Hong Kong are available from the Hong Kong Census and Statistics Department. For other countries, quarterly GDP data are available from DataStream.
In Table 2, we show the summary statistics of quarterly data on democratic accountability and GDP growth rate for the eight economies over the sample period. The average rating of democratic accountability over the sample period has the highest value for the Philippines and the lowest for Hong Kong. Similar to other categories of political risk, democratic accountability is also remarkably stable over time, as shown by the low standard deviations in Table 2. South Korea and Indonesia are the two economies that have experienced the highest volatile movements in democratic accountability. As for the quarterly GDP growth rates, the standard deviations are high for almost all HAPEs (with the possible exception of Taiwan), which reflects the volatile growth experience of HPAEs, such as during the 1997 Asian financial crisis.

Econometric Method
We adopt the autoregressive distributed lag (ARDL) bounds test methodology of Pesaran et al. (2001) to test for the existence of a longrun relationship between democracy and GDP growth using HPAE data from 1984 to 2003. This technique does not require the researcher to assume that the underlying democracy and economic growth series are I(0) or I(1). In particular, we find difficulty in ascertaining whether the index of democracy is stationary. On the one hand, this index can be labeled as stationary because it can only take on a limited range of discrete values. On the other hand, it clearly exhibits patterns of nonstationarity in formal unit root tests.^{6} Thus, using the ARDL bounds test approach is especially appealing in this context to avoid confronting the problem of identifying the order of integration for the index of democracy.
The ARDL regression yields a test statistic that can be compared to two asymptotic critical values. If the test statistic is above the upper critical value, then the null hypothesis of no longrun equilibrium relationship between democracy and economic growth can be rejected regardless of whether the series are integrated of the order of zero or one. Alternatively, if the test statistic falls below the lower critical value, then the null hypothesis cannot be rejected, again regardless of whether the series are I(0) or I(1). If the test statistic falls between the bounds of the two critical values, then the result is inconclusive.
The ARDL bounds test approach begins with an unrestricted VAR in levels:^{7}
(1) ,
where . Here and are the index of democratic accountability and the growth rate of GDP at time . As noted earlier, the two series, and , can be either I(0) or I(1). is a vector of constant terms, , and is a matrix of VAR parameters for lag . The vector of error terms , where is positive definite and given by
(2) .
Given (2), can be expressed in terms of as
(3) ,
where .
We manipulate the VAR model of (1) to obtain a vector error correction model (VECM), such as
(4) ,
where , L is the lag operator, and
(5) .
in (4) is the longrun multiplier matrix and is given by
(6) ,
where is a 2 x 2 identity matrix. As each of the series can be either I(0) or I(1), the diagonal elements of the matrix are left unrestricted. Moreover, we can only test at most one longrun relationship under this procedure. Hence, a zero restriction on one of the offdiagonals of the matrix is required. We impose , which implies that there is no feedback from to the level of . Using the terminology of Pesaran et al. (2001), the growth rate of GDP is a longrun forcing for democracy. However, the validity of the forcing variable assumption can be tested for the exclusion of the lagged democratic accountability in the GDP growth equation of the vector error correction model (VECM) that was described by (4).
Given the assumption of and (3), the equation for democratic accountability from the VECM of (4) can be written as:
(7) ,
where , , , , and . We can interpret (7) as an ARDL(p, q) model, where p is the number of lagged differences of the growth of GDP and q is the number of lagged differences of democratic accountability. In practice, p and q do not have to be the same, and our search for optimal orders of p and q is based on two considerations. The optimal ARDL(p, q) model must be parsimonious and it must be free of serialcorrelation.
In (7), the null hypothesis of no longrun relationship between democratic accountability and the growth rate of GDP is . We first estimate (7) by OLS and then calculate the Fstatistic for the null hypothesis of against the alternative hypothesis that and . The distribution of the test statistic depends on the order of integration of the two underlying series, and Pesaran et al. (2001) provide the critical values that are calculated from the distribution of test statistics under the null hypothesis. We accept the null hypothesis of no longrun relationship between democratic accountability and the growth rate of GDP if the test statistic falls below the lower critical value. We reject the null hypothesis in favor of the alternative hypothesis if the test statistic exceeds the upper critical value, regardless of whether democratic accountability or the growth rate of GDP is I(0) or I(1). If the test statistic falls between the lower and upper critical values, then the result is inconclusive.
Under the alternative hypothesis that both and in (7), there is a stable longrun relationship between democratic accountability and the growth rate of GDP, which is described by
(8) ,
where , , and is a meanzero stationary process. Once (7) is estimated and a longrun stable relationship is detected, we can use (8) to calculate the longrun equilibrium relationship.

