Financial integration and economic performance of the SACU countries: do different types of capital

1
Financial integration and economic performance of
the SACU countries do different types of capital
stocks matter?

• By
• Prof. Meshach Aziakpono
• Department of Economics, Rhodes University
• Prof. Philippe Burger
• Department of Economics, University of Free State
• And
• Prof. Stan du Plessis
• Department of Economics, University of
  Stellenbosch
• Paper for presentation at the 16th Annual
  Conference of the Southern African Finance
  Association (SAFA), January 14-16, 2009

2
Outline of Presentation

• Introduction
• Objective of the Study
• Definition and measurement of FI
• Theoretical Framework
• Empirical Literature
• Methodology
• Data
• Results
• Conclusion

3
1. Introduction

• Increasing integration of financial markets
  across international borders is a remarkable
  feature of the last two decades.
• Long history of official integration arrangement
  among SACU countries

4
2. Objective of study

• To provides evidence on the FI-growth nexus based
  on the experience of the SACU countries.
  Specifically
• To examine the nature (i.e. the sign and size) of
  the effects of FI on the level of per capita
  income in each of the SACU countries. As a
  corollary, to explore the effect of economic
  performance on FI in SACU.
• To explore the direction of causality between the
  level of income per capita and FI in each of the
  SACU countries.
• To investigate the relative effects of the
  different types of capital stock on economic
  performance in the SACU countries.
• To explore the channels through which FI affects
  economic performance of each of the economies.
• Lastly, to determine whether or not the results
  of the above analysis are robust to controlling
  for other possible determinants of economic
  performance and FI.

5
3.Definition and measurement of FI-1

• Definition
• Economic literature does not offer a unique
  definition of financial integration
• Terms such as financial openness, external
  financial liberalisation, financial globalisation
  and capital account liberalisation have also been
  used in connection to financial integration
• De jure (prerequisites for financial integration)
• Removal of any administrative and market-based
  restrictions on capital movement across borders
• Removal of regulatory, legal and tax
  discrimination between foreign and domestic
  suppliers of financial services
• Existence of a transparent and efficient legal
  system and quality law enforcement
• De facto financial integration (integration
  outcomes)
• Relates to price- or return on assets, capital
  flows, household decisions, and corporate policy
  

6
Figure 1 Framework for Categorising Measures of
Financial Integration
7
4. Theoretical Framework Role of FD
8
4.Theoretical Framework Role of FI
9
4.Theoretical Framework Channels of FI

• Direct channel
• through increasing domestic investment by
  augmenting domestic saving and reducing the cost
  of capital
• Because of international risk sharing
• Indirect channels
• simulation of domestic financial sector
  development,
• transfer of technology and managerial know-how,
• macroeconomic discipline and
• signaling effect

10
4.Theoretical Framework Potential costs of FI

• Potential costs
• geographical concentration of capital flows
• lack of access to such capital flows by small
  countries,
• domestic misallocation of capital flows,
• loss of macroeconomic stability,
• pro-cyclicality of short-term flows,
• volatility of capital flows and
• risk of entry by foreign banks.

11
5. Empirical Literature-1

• The burgeoning and increasingly innovative
  empirical literature provides conflicting
  evidence on the effects of IFI on economic growth
  and welfare
• From a methodology perspective studies range
  from
• the use of simple ordinary least squares (OLS)
  cross-sectional regression analyses (cf. Klein,
  2003 Arteta et al. 2001 Edwards, 2001 Klein
  and Olivei, 1999 Kraay, 1998 Rodrik, 1998
  Quinn, 1997),
• advanced dynamic panel estimation techniques that
  control for potential biases in the earlier
  methods (cf. Kim et al. 2005 Laureti and
  Postiglione, 2005 Calderon et al. 2004 Edison
  et al. 2002 Reisen and Soto, 2001)
• time series studies that employ the vector
  autoregressive (VAR) model (cf. Jin 2006 Kim et
  al. 2004).

12
5. Empirical Literature-2

• Recent studies have examined a range of other
  issues in addition to investigating the standard
  question of whether or not FI really promotes
  economic growth. These include
• the conditions under which FI contributes to
  economic growth, e.g.
• institutional quality (cf. Klein, 2005, Edison et
  al. 2004 Edison et al. 2002),
• level of per capita income (cf. Klein, 2007
  2003 Edison et al. 2004 Edison, et al. 2002),
• ethnic and linguistic heterogeneity (cf. Chanda,
  2005)
• whether the effects of FI differ during different
  eras of financial globalisation (cf. Schularick
  and Steger 2007 2006)
• whether the effects of FI depend on the type of
  capital inflows (cf. Laureti and Postiglione,
  2005 Collins, 2004) and
• the sources of effects of integration (cf.
  Schularick and Steger, 2007 Collins, 2004).

13
5. Empirical Literature-3

• Authoritative surveys of the literature by other
  authors, such as Kose et al. (2006), Collins
  (2004), Edison et al. (2004) and Prasad et al.
  (2004) found mixed evidence.
• For instance, Kose et al. (2006) found that most
  of the studies obtain no effect or the results
  were ambiguous (results that are not robust
  across alternative specifications) for developing
  countries.
• Hence, they concluded that if financial
  integration has a positive effect on growth, it
  is apparently not robust, once the usual
  determinants of growth are controlled for
  (p.13).

14
5. Empirical Literature-4

• An analysis of many of the previous studies on
  the effects of FI on growth reveals that the
  mixed results reflect differences across the
  studies
• the measures of FI used,
• country sample,
• time period cover,
• econometric methodology, and
• the set of control variables used.

15
5. Empirical Literature-5

• A major weakness of most previous studies relates
  to the frameworks used, such as the
  cross-sectional and panel data approaches
• the inability of the approaches to cater
  adequately for the endogeneity of FI and some of
  the right-hand side variables
• hence they cannot be used to establish causal
  relation between economic growth and such
  variables.
• the coefficient estimates are only average
  effects for a sample of countries covered and do
  not represent any particular country in the
  sample. This makes it difficult to draw
  country-specific policies conclusions based on
  such results.

16
6. Methodology-1

• To achieve the above objectives, the analysis in
  this study draws on previous studies in a number
  of ways
• First, following the trend in recent literature
  (IMF, 2007 Collins, 2004 ODonnell, 2001
  Reisen and Soto, 2001 Kraay, 1998), the current
  analysis employs different de facto measures of
  FI as opposed to the de jure measures.
• Second, it uses the standard control variables
  used in growth literature.
• It departs from most of the previous studies that
  use cross-sectional and panel data approaches by
  adopting a time-series country-specific approach.
• Hence, it is similar to Kim et al. (2004) and Jin
  (2006). But, unlike the Kim et al. (2004) and Jin
  (2006) studies, the current study explicitly
  explores the endogeneity of each of the variables
  used as well as models the long-run causality
  between FI and per capita income
• Moreover, it focuses on SACU countries

17
6. Methodology-2

• The study adopts a trivariate cointegration and
  error correction modelling technique based on the
  Johansen maximum likelihood approach
• Model
  (4)
• where, LPY, FI and CV represent the level of per
  capita income, FI and a control variable.
• Unit root test DF-GLS and Ng and Peron 2001
• Cointegration test
• weak exogeneity test (causality test)

18
6. Methodology-3

• Due to the low degree of freedom due to the small
  sample size, the study follows Kim et al.
  (2004630) in including only three variables in
  each model (i.e. a trivariate model).
• By sequentially adding and subtracting a control
  variable, it is possible to test the robustness
  of the long-run relationship obtained in the
  models, i.e. whether or not the long-run effect
  is dependent on the control variables
• The quality of models where cointegration is
  found and the variable of interest (LPY or FI) is
  endogenous in the mode is evaluated using two
  criteria
• First, must pass the residual diagnostic tests
  serial correlation and heteroskedasticity tests
• Second, a threshold of 30 explanatory power
  (the adjusted R2) for any model to be reported
  for analysis

19
6. Methodology Econometric procedure-1
20
6. Methodology Econometric procedure-2
21
6. Methodology Econometric procedure-3
22
7. Data

• In line with standard practice in most
  time-series studies (cf. Demetriades and Hussein,
  1996 Arestis and Demetriades, 1997 Luintel and
  Khan, 1999) this study uses the level of per
  capita real GDP as a measure of economic
  performance
• We four measures of IFILFDIL, LDL, LFIA and LFIL
  
• We use as control variable
• Two measures of financial development (LFDC and
  LFDL)
• the ratio of domestic investment to GDP (INV),
• the ratio of government expenditure to GDP (GE),
• inflation rate (INF),
• trade openness measured as the ratio of sum of
  export plus import to GDP (OPN)
• Data covers the period 1970 to 2004, although,
  for many series, data was not available for the
  entire period. The series ranges from 29 to 34
  continuous annual observations
• Sources dataset on External Wealth of Nations
  Mark II by Lane and Milesi-Ferretti (2006), IMF
  IFS 2007 CD-ROM , World Bank World Indicators

23
8. Results

• Altogether, the study estimates 25 models each
  for Botswana, South Africa and Swaziland. Due to
  data limitation, the study estimates only ten
  models for Lesotho.
• Of the 25 models estimated, a stable long-run
  relationship is found in 18 in Botswana, 13 in
  South Africa and 5 in Swaziland. In Lesotho, 7 of
  the 10 models estimated produce a stable long-run
  relationship.

24
Botswana-1
25
Lesotho-1
26
South Africa-1
27
Swaziland-1
28
Botswana-2
29
Lesotho-2
30
South Africa-2
31
Swaziland-2
32
8. Results Botswana

• To summarise the results of the analysis for
  Botswana,
• Overall the results do not produce robust
  evidence on the effect of FI on output
• The evidence does not support the contention that
  the stock of FDI has a stronger positive effect
  on economic growth than the stock of debt.
• The results suggest that FI affect the level of
  output through both the direct channel
  (investment channel) and indirect channels such
  as through domestic financial system development.
  
• But such direct and indirect effects of FI on
  output level appear to be negative in Botswana.

33
7. Results Lesotho

• In summary, the results highlight a number of key
  findings for Lesotho
• Overall, the results suggest that an accumulation
  of foreign liability of banks causes an increase
  in the level of per capita income. But the effect
  is not conclusive if banks accumulate foreign
  assets.
• The evidence suggest that FI affects the level of
  income through both the direct channel
  (investment channel) and the indirect channels,
  but while the direct channel appears to have a
  positive effect, the indirect channel through
  deposit market development has a negative effect

34
7. Results South Africa

• In summary, the following are the key findings
  for South Africa.
• Overall, the effect of FI on output depends on
  the types of stock of capital. Whereas, the
  FDI-related capital cause output to increase, the
  debt related flows have a negative effect on
  output, thereby supporting the view that FDI
  inflows are better than debt inflows in promoting
  economic growth in South Africa.
• The results suggest that FI operates through both
  the direct and indirect channels, with the
  investment channel having a strong positive
  effect on the level of output in South Africa.
• The results also suggest that FI operates through
  other indirect channels, such as domestic
  institutional development, signalling effect,
  with such channels appearing to have positive
  effect on output in South Africa.

35
7. Results Swaziland

• In summary, the weight of evidence strongly
  suggests a positive effect of FI on output levels
  in Swaziland.
• Notably, the positive effect stems from the
  accumulation of debt-based capital stocks, while
  FDI inflows has a negative effect in Swaziland.
• Financial integration seems to operate through
  both the direct and the indirect channels.
• Also the result suggests that an increase in the
  level of income leads to an increase in the
  degree of FI, specifically accumulation of bank
  foreign assets.

36
8. Conclusion

• The effects of FI on the output level, the
  results are mixed the effects vary from country
  to country and depend on the types of capital
  stock.
• The effect of FDI is negative in Botswana,
  positive in South Africa but ambiguous in
  Swaziland.
• The stock of debt liabilities has a negative
  effect in Botswana and South Africa while the
  effect is positive in Swaziland.
• The ratio of foreign assets of banks has an
  ambiguous effect in Botswana, Lesotho and South
  Africa while no effect was detected in Swaziland.
  
• Lastly, the ratio of foreign liabilities of banks
  has a positive effect in Lesotho and Swaziland
  and a negative effect in South Africa, while the
  effect is ambiguous in Botswana.