The effect of financial crisis 2008 on firm productivity

[Jahangir Alam]

I do not know this is the right place to ask.

I have a firm-level panel data for a country in Europe. I would like to estimate the effect of financial crisis 2008 on firm productivity. I would like to use difference in difference method but I could not think any control country for this type of estimation.

Is there any econometric method or statistical method for estimate the effect of financial crisis 2008 on firm productivity?

Please write your thought!

Thanks in advance!

[Carlo Lazzaro]

Jahangir:
what does the literature in your research field report about that topic?

[Erik Tollefson]

One group of scholars at the Bank of England perform a pre-post examination to assess the impact of the financial crisis on producitvity in England (https://www.ifs.org.uk/uploads/Prese...915/RRiley.pdf). Due to differences in firm structure and productitivty it might make sense, if you have the data, to perform a similar analysis of firm productivity. They take data from 1999-2007 for the pre-period, data from 2008-2009 for the financial crisis, and data from 2010-2013 for the post-period.

[Clyde Schechter]

This is way outside my domain of content expertise. But it would surprise me to find control countries that didn't experience the financial crisis of 2008, except perhaps countries that are economically undeveloped, and so would not be comparable for that reason. But the financial crisis played out differently in different countries. If you can find a suitable measure of the intensity of the crisis in different countries, you might reconceptualize the problem as modeling the relationship between your outcome and crisis intensity.

[Phil Bromiley]

If you don't have countries without the event, then you have difficulty doing a conventional difference in difference, although Clyde's approach is an interesting alternative. If I were following Clyde's suggestion, I'd look to see if anyone has done something similar in another research area with difference-in-difference.

If you give up on difference in difference, then you can easily estimate either a difference in means, or a regression with a dummy for the period after the crisis.

[Dimitriy V. Masterov]

Cross-posted at CV. Note the policy on this.

[Jahangir Alam]

@Erik Tollefson Thanks for your idea and link. I am thinking to make a causal statement. Any other idea?

@Carlo Lazzaro I could not find any study with my limited knowledge that make a causal statement.

[Jahangir Alam]

@Clyde Schechter Thanks. I like your idea about intensity. Does 'energy consumption by industries' capture the intensity of the financial crisis? Could I compare highly affected country and less affected country?

@Phil Bromiley Currently, I am comparing means before and after. Thanks both of you.

[Clyde Schechter]

@Clyde Schechter Thanks. I like your idea about intensity. Does 'energy consumption by industries' capture the intensity of the financial crisis?

That's an example of what I meant by my caveat that I'm way out of my domain of content expertise here. I'm an epidemiologist. I don't know if energy consumption would be a reasonable measure of the intensity of the financial crisis. From a common sense perspective, it seems reasonable to me, but my knowledge of economics/finance is quite limited. This is a question I would urge you to take up with people in your own discipline. (Or, as there are several such on this Forum, perhaps some of them will respond here.)

The reason I suggested the approach is that in epidemiology we sometimes encounter situations where we want to study whether a putative risk factor actually enhances risk of a certain disease (or perhaps a putative preventive factor reduces it), but it is difficult or impossible to find adequate numbers (sometimes not even any) unexposed controls. So in that situation we sometimes do a dose-response study. I was reasoning by analogy to that.

[Dimitriy V. Masterov]

Energy consumption seems more like an intermediate outcome or an input for productivity, if by productivity you mean total output in some industry for which electricity is an input. If you mean that the financial crisis makes people forget how to make things, so they produce fewer widgets given the same inputs, maybe that is more reasonable.

[Jahangir Alam]

@ Phil Bromiley Could I compare highly affected country and less affected country? Explaining relative coefficient using difference in difference method. Could we do like this?