FINANCE AND THE ECONOMY

Finance and The Economy

Finance and The Economy

Part I

 

1997

2000

2001

2002

2003

2004

2005

2006

2007

2008

Currency & Deposits

38

32.3

30.3

24

20.4

23.3

27.3

27.1

27.1

25.9

Securities Other Than Shares

6.7

4

1.7

2.1

1.5

1.6

1.7

1.6

1.6

1.3

Insurance Technical reserves

35

45.1

44.5

50.8

52.4

53.6

52.9

52.7

52

53.3

Other Accounts Receivables

6.5

5.9

4.5

3.4

2.7

3

3.2

3.1

3.1

3.2

Net Acquisition of Financial Assets

86.2

87.3

81

80.3

77

81.5

85.1

84.5

83.8

83.7

Loans

0.9

0.3

0.5

0.3

0.2

0.2

0.3

0.2

0.2

0

Shares & Other Equites

12.9

12.4

18.6

19.4

22.8

18.4

14.7

15.3

15.9

16.3

Net Acquisition of Financial Liabilities

13.8

12.7

19.1

19.7

23

18.6

15

15.5

16.1

16.3

Part II

Following Islam (1995) we used the time dimension of the entire panel data to carry out exercises of absolute ß-convergence. An important technical issue in switching from a single cross section to a panel data framework is the choice of the time span dividing the total time range. We picked the value of 5 years. This seems a reasonable trade off between the competing needs of increasing the available observations and reducing short term disturbances. The value of 5 years appears to be less sensitive to the business cycle IFC Bulletin No 31 389 fluctuations and serial correlations in the residuals.10 Given the temporal extension of our dataset, this choice produces a total of 2005-1980+1=26/5=5 observations for each of the countries. The results are robust to a different choice of the time interval (see the following of this subsection for further details). A lower frequency (less than 1 observation every 5 years) would yield too few observations. We run three different estimation methods against our available dataset: pooled OLS, random and fixed effects. Pooled OLS is the easiest method from the computational standpoint. OLS technique provides a simple benchmark but the estimated ß coefficient is inconsistent when the unobservable country effects are correlated with other explanatory variables. For this reason we also add the random-effects estimator that introduces a random country specific element. With this method, when the unobservable country specific component is uncorrelated with the regressors, we have consistency for the estimated ß coefficient. Aiming also at efficiency we finally flanked our results with the fixed-effect estimation which includes a country specific constant mirroring all the unmeasured country effects. The use of different estimation methods provide us with a sound robustness check (see Cellini, 1997 and Islam 1995 for the same approach).11 Gross disposable income is used as a scale variable.

While banking disintermediation is common to all OECD countries, the importance of deposits still remains different. This result is similar to the evidence reported in Di Giacinto and Esposito (2006). Also other studies found that the weight of safe assets in household portfolios differs across countries. This result may be influenced by national peculiarities, such as fiscal treatment of deposits, the forms of competition between banks and other financial intermediaries, the different weight of the Post office, and regulatory and institutional factors that influence the offer of deposits. Table 3 refers to insurance products. The ß-coefficients are always negative and statistically significant. Countries where public pension schemes were relevant in the past followed the example of financial systems where private insurance and pension products were traditionally more common among households. We may refer to the catching up of Spain and Italy (see figure 2). Population ageing is a trend common to all industrialised countries and leading to important restructuring of the financial industry, such as a growing role ...