Dear Gitz Negi

Greetings,

Please find the attached file about your topic may be useful for you.

Best Regards

Obviously the introduction of Banking crisis as a variable distorts the effects of financial development variables mainly due to multi collinearity effects . You can check up the correlation matrix  between these variables . Ideally that control variable should have improved the results but the opposite effect - may be due to collinearity effects .

I probably do not have the necessary time to give the detailed answer this question deserves, but it should importantly be the following: 

( This is also a key prioritisation in 1 as most important , to 5 and 6 as also contributing causes/methodology resolution) 

1. The Central Banks have to prioritise liquidity in the Banking markets to keep credit flowing, and the eased liquidity supports growth of speculation yet alsop tremendously contributes to a happy consumer index , consumption and growth. 

2. Governments support Industry and trade. Trade contracts get tremendously simple during a crises. This excludes pockets of degrowth like Europe early on and China now. ( and the continuing development shows the detriment of these pockets is not the be all end  all of economic growth determinants

3. The statistical study employed suffers from wrong definition of crises periods and the wrong use /non use of dummy variables. Annual datas always show robust effects across decades and years for statistical reasons and they are not good indicators of causality ( You cannot use the coefficients) most importantly because of endogenity (the crisis effects the trade that effects the crises, critically the exchange rate movements as a correcting factor should be controlled to find the effect on trade or developed - developed relations only chosen. 

4. In your methodology, intuitively the latter regression study should be more effective statistically. However, the important thing is to measure the qualitative aspects and use them as verification of the results. 

5. The presence of Emerging Countries presents easy opportunities to continue economic development as exports. US centric data will also show economic development and growth from continuing consumption even as car use degrows and median salaries are affected negatively. That should affect what studies you conduct to verify your result. (use different FRED series ) 

6. Use of Fama macbeth cross sectional regressions and/or performance matched samples across an event study design / other matching pre/post designs should also compete as good soolutions

@ Faris Alshubiri ....thanks for your reference.

@M.Thomas Paul .....Thanks for your reply.

Actually when i used banking crisis dummy as a control, variable results were unaffected. In second regression i used interaction term (Financial development*crisis dummy) and kept both variables too. Coefficients for interaction terms were significant but positive. It implies that financial sector has positive impact on growth during crisis which is quite unconventional norm.

@Amit Mittal.....Thank you very much for such a helpful reply. Could you please further comment on the following.

Since my data set (1999-2012) consists high, upper and low income group economies. And 75% of data belongs to high income group. So the vanishing effects were quite visible in results. But interaction with crisis is saying a different story by giving positive coefficient.

As explained by Amit Mittal    during financial and banking crisis , some positive policy  variables  positively impact the economy :    the high liquidity and low interest rate environment  provided by the Central Bank of the country, tighter banking regulations along  with bailing out the banks , the depreciation of the currency to promote exports, and the economy stimulating packages by the governments  .These positives impacts can sometimes overtake the negative impacts of moral hazards , deleveraging  by the banks and firms, financial market volatility  etc.

Can I ask exactly what is your measure of "financial development"? If it is credit on GDP, I would check if it is performing credit or credit net of provisions for non performing loans ... sure those increase during the crisis!!!! 

Interactions in a crisis will give different responses in developed - developed, developed - emerging and emerging - emerging as well as uphill flows from emerging - developed. or in your case, the nine different sets between high-high, high-medium, high-low, ...low-low. Currency movements were initially assumed to be the major reason as observed in the literature of eichengreen and found insignificant(1999) , , eichengreen has various updated reviews and data analysis till 2009 and 2010 with the various dimensions, Independently each of the dimensions of currency moves etc are found to be insignificant. Also read his text on Hall of Mirrors by Eichengreen, Demirguc and Detriagache, Reinhart and Rogoff .

Each however prove that for specific causation from the crisis evidence is available as crisis time response is different

@Amit Mittal ....Thanks a ton for your help. Your references really helped a lot.

@M.Thomas Paul ....Thank you very much for your help. 

could you please help me with one more doubt. I really do not know how to justify positive effect of interaction term(Private credit*banking crisis) as an independent variable and negative effect of banking crisis dummy as a control variable in same regression equation? 

Is it correct to say that the overall impact of banking crisis is negative on economic growth but private credit during banking crisis is supporting economic growth through various other channels? 

Regards.

@Alessandra Dal Colle .....Thanks for replying. 

Measures are Private credit to GDP, Broad money, stock turnover ratio, stock traded and stock market capitalization.

Regards.

 I have the same results when I include the banking crisis variable. I uses all measures of financial development (Private credit to GDp, Broad money etc...). However, The coefficients of financial development variables are not significant. The banking crisis variable are significant.