3 Investment, credit and output growth - a VAR analysis
clearly be better, but jointly, they give a more complete picture of the link between financial
development and growth.
Generalized impulse response functions
Figure 2 reports the generalized impulse responses from our first VAR that includes the
variables net domestic product, investment and bank lending. Our main interest is in the
effect that banks have on the net domestic product, which is displayed in Panel A. There,
a statistically significant effect for about four years exists. Panel B shows that there is in
addition another indirect effect. For a period of three to four years, an unexpected increase in
bank lending increases investment. It is well known that investment, in turn, has a positive
• л’тлтл 1 9
impact on NDP.12
Figure 2: Generalized Impulse Responses for Net Domestic Product, Investment and Bank Lending
1 2 3 4 5 6 7 8 9 10
Panel A |
________________________Panel B_________________________ | |
— |
Reaction of NDP |
Reaction of I |
_ |
___to a shock in B__________________ |
to shock in B |
Note: The solid lines trace the impulse responses of net domestic product
(NDP) to shocks in investment (I) and bank lending (B) for the years 1870
to 1912.
Table 3: Variance Decomposition for Net Domestic Product, Investment and Bank Lending
л r
Years
Variance Decomposition 5 10
NDP variance due to B (in percent) 24.009 23.129
[12.374] [12.294]
I variance due to B (in percent) 30.006 29.281
_____________________________________________[12.470] [12.541]
Note: The variance decomposition of the forecast error is
shown for the three-variable VAR, including net domestic
product (NDP), investment (I) and bank lending (B) for the
years 1870 to 1912. The values in parentheses indicate the
standard deviation.
Although the impulse response functions have revealed a clear link between aggregate bank
credit and net domestic product, they do not allow to assess the importance of these shocks
12Indeed, the impulse response for NDP and investment reveal a positive but short-lived impact on NDP, when
investment is shocked unexpectedly. Because this effect is often reported in the literature, we do not show
this graph in this paper.