Skip to content

Testing Creative Destruction in an Opening Economy: the Case of the South African Manufacturing Industries

Literature › Works

Philippe Aghion; Johannes Fedderke; Peter Howitt; Chandana Kularatne; Nicola Viegi (2008). Testing Creative Destruction in an Opening Economy: the Case of the South African Manufacturing Industries.
Type working paper
Scope Post-1994 South African economy
Method econometric
Period studied 1988–2002
Regions South Africa
Themes manufacturing industrial policy, transition post apartheid, comparative theory, methods data sources, business and state
Citation key aghion2008

Argument

The paper tests a Schumpeterian model (Aghion and Howitt 2007) in which trade liberalization affects productivity growth directly and indirectly. The indirect channels are a scale effect, which favours smaller sectors, and a backwardness effect, which depends on a sector's distance from the technological frontier. It estimates fixed-effects and GMM panels for 23 South African 3-digit manufacturing sectors, with the US as the frontier. Smaller sectors gain most from liberalization, and South African sectors are not far enough behind the frontier for liberalization to depress productivity growth. Pricing power and industry concentration are strongly negatively associated with productivity growth, and exchange-rate depreciation has only a very small effect. The authors conclude that the net effect of liberalization is positive and that competition policy and liberalization are preferable to depreciation as growth instruments.

Evidence and method

UNIDO International Industry Statistics 2004 (used to compute US TFP), the TIPS South African 3-digit manufacturing database (1970-2004), Edwards (2005) openness measures (effective protection, tariffs, export taxes, anti-export bias, 1988-2003), mark-ups from the Roeger method (Fedderke et al.), and Rosenbluth concentration indices. The panel is about 23 sectors over 1988-2002. The authors flag data-quality problems after 1996.

Findings

  • Scale of production has a strong positive effect on productivity growth, with a coefficient of about 0.45-0.70. (p. 12)

    if the ratio of South African to US value added production rises by one percentage point, productivity growth in South African manufacturing would rise by approximately half a percentage point

  • Smaller sectors benefit more from liberalization, as the scale-by-protection interaction is negative. (p. 13)

    smaller sectors stand to gain more from an opening of the trade regime in scale of production terms than do large sectors.

  • South African sectors are not far enough behind the frontier for liberalization to reduce productivity growth. (p. 13)

    from the technological frontier, for trade liberalization to exercise a negative impact on productivity growth

  • Pricing power (Lerner index) strongly depresses productivity growth, and the effect is larger once trade protection is controlled for. (p. 13)

    unit increase in the Lerner index now results in a 2 to 3.5 percentage point loss in productivity growth,

  • Industry concentration (Rosenbluth index) is strongly negatively related to productivity growth. (p. 13)

    a 0.01 unit increase in the concentration index reduces productivity growth in South African manufacturing by between 0.77 and 0.88 percentage points

  • The direct effect of liberalization on productivity growth appears negative (protection coefficient positive in the full specification), though the net effect including indirect channels is positive. (p. 15)

    Importantly, while the direct impact of trade liberalization on productivity growth appears to be negative

  • The exchange-rate depreciation effect is economically trivial. (p. 14)

    for a 1 Rand to the Dollar depreciation, the gain in productivity growth amounts to 0.01 percentage points

  • Liberalization was substantial by Edwards' measure (unweighted effective protection) but contested and incomplete in other measures. (p. 8)

    protection in manufacturing fell from 48% to 12.7%

  • Fedderke and Vaze find South African industries remain heavily protected, with protection rising in some industries. (p. 7)

    South African industries are still heavily protected. In some industries, protection appears to have increased (Fedderke and Vaze (2001).

  • The authors flag the reliability of post-1996 South African industry data as a limitation. (p. 8)

    There are questions over the reliability of South African industry data post-1996.