Introduction to Empirical Section



Download 488.91 Kb.
Page5/11
Date25.02.2021
Size488.91 Kb.
1   2   3   4   5   6   7   8   9   10   11

Explanatory variables


The domestic explanatory variables of interest are the level of democracy and economic policies. For democracy, I use the measure developed by Przeworski et al 2001 (and updated by Cheibub and Gandhi 2004). This indicator is a dummy, which in its reversed form is 1 for democracies and 0 for non-democracies. The decisive factor for this measure is the institutional factor of primary interest for this overall project, i.e., in democracies the government leader risks electoral overthrow, whereas in non-democracies he doesn’t. Because a leader may win free and fair elections, the ACLP democracy measure is coded retroactively for democratization – when a leader who was a dictator loses an election, the regime is coded as a democracy back to the date when the new electoral rules were established. For example, J.J. Rawlings of Ghana ruled without elections until the end of 1992, and then allowed for multi-party elections, which he won in late 1992 and 1996. In 2000, however, his chosen successor, Atta Mills, was defeated in an election. Ghana is therefore coded as a democracy beginning in 1993, when multiparty elections started. Although the coding in such cases depends on future events, many observers considered the election in Ghana to be free and fair in 1992 and 1996, and democracy measures that rely on expert opinion such as FreedomHouse and Polity also date Ghana’s democratization to 1992. The correlation among the three measures of democracy ranges from 82-86%.

Because no single measure is ideal to differentiate between “capitalist” and “statist” investment policies,10 I have created a summary measure, similar to the approach of Sachs and Warner (1995) in creating a trade openness dummy. There are two policy areas of importance in investment policies when comparing statist to capitalist policies. First, statist policies restrict international capital flows, including foreign investment in private enterprises. Second, statist policies invest in state-owned enterprises and nationalize existing foreign-owned enterprises. Capitalist policies, on the other hand, allow for the free flow capital, do not invest in new state-owned enterprises, and divest existing state-owned enterprises.

I therefore code economic policy as statist if they follow any of the following criteria:


  1. The Financial Openness Index (Brune 2004) index is 0, meaning all forms of capital inflows and outflows are restricted (FOI).

  2. Public Investment (which includes state enterprise investment and government fixed investment, Pfefferman et al 1999) is greater than 10% of GDP (SPI).

  3. Major state enterprise investment or private enterprise nationalization occurs in more sectors than does state enterprise privatization or liquidation (SEI). 11

Otherwise, economic policy is coded as capitalist. See Appendix 1 for a further explanation of each component in the economic policy variable.



Most countries in sub-Saharan Africa pursued statist economic policies in the 1970s until the debt crisis. The debt crisis also coincided with some democratization, but these democracies were short-lived. The share of countries pursuing capitalist polices increased in the 1980s, leveled off for several years, then rose dramatically again at the end of the 1990s. The number of democracies increased during the early part of the 1990s, stalled for a few years, then rose again into the turn of the century.


In addition to the explanatory variables of interest, I also include the following control variables:

The total volume of net inflows (in millions of US 2000 dollars) to sub-Saharan Africa for each category in a given year proxies for the various push factors such as interest rates, economic growth, and herd behavior in the developed world.

Domestic control variables come from the capital flows literature. (All economic data come from the World Bank’s World Development Indicators unless indicated otherwise.) I begin with four of the determinants found to be the most robust in predicting FDI inflows in an Extreme Bound Analysis (Chakrabarti 2001): 12

Income level (GDP per capita) has the most robust, and positive, effect on FDI, since a more developed economy can absorb more FDI, and wealthier consumers can purchase more products from market-seeking FDI. (Since much of the FDI to Africa is resource-seeking, however, the effect of this factor might be attenuated.) Furthermore, low income levels are correlated with capital scarcity, and standard economic models expect capital to flow from capital abundant to capital scarce economies.13

Trade Openness (exports plus imports as a percentage of GDP) has the second most robust, and positive, effect on FDI. Although FDI can serve as a substitute for trade, and the literature discusses “tariff-jumping FDI,” trade openness generally has a positive effect on FDI. Easy access to imports lowers the cost of doing business for foreign subsidiaries, and the ability to export is attractive to efficiency-seeking FDI.

Economic growth (percentage change in GDP) is an indicator of high productivity and future profit opportunities, and therefore should attract FDI.

Net exports (exports minus imports as a percentage of GDP) is a proxy for the current account balance. Negative net exports (or a current account deficit) tend to be correlated with positive capital inflows. When imports exceed exports, an excess of currency will be held in foreigners’ hands, and one use for that currency is to send it back as capital flows.

Where FDI inflows are the dependent variable, I also include the stock of FDI per capita, whose expected effects are ambiguous.14 On the one hand, accumulated FDI should decrease the need for additional FDI (and higher levels of FDI stock enable higher levels of reversed flows). On the other hand, there may be agglomeration effects, whereby existing FDI makes new inflows of FDI more productive. Agglomeration effects, however, generally occur in knowledge-intensive industries, which are less prevalent in Africa.

Since much of the capital flowing to sub-Saharan Africa is resource-seeking, I also include a measure of natural resource endowment. I use the sum of the value of mineral and ore production and energy production (oil, natural gas, and coal).

When portfolio equity flows are the dependent variable, I also include a dummy indicating the presence of a stock market.15 Although FDI is long-term, portfolio equity is short-term, and therefore more likely to consider short-term risks such as volatility. I use the standard deviation of (official) exchange rates over the previous ten years to measure exchange rate risk. I also include a dummy for South Africa, which received the lion’s share of portfolio investment.16

Lenders are influenced by the perceived ability of the borrower to repay loans. African governments and firms must repay loans in foreign currency, which is earned with exports. I therefore measure the perceived ability to repay loans with the total debt service (TDS) as a percentage of exports. (The need for foreign currency to repay loans is also a reason that natural resource endowments are important to lenders).

Data permitting, the analysis includes observations from 34 countries covering the years 1972-2003. I use ordinary least squares (OLS) with panel-corrected standard errors (PCSE), as recommended by Beck and Katz (1995). Because the number of years is roughly equal to the number of countries, techniques such as FGLS are inappropriate. 17 I include the lagged dependent variable, which also serves as a proxy for any omitted variables. All explanatory variables are lagged one year to address endogeneity concerns unless otherwise noted.





Share with your friends:
1   2   3   4   5   6   7   8   9   10   11




The database is protected by copyright ©essaydocs.org 2020
send message

    Main page