RIETI - Hitotsubashi - Gakushuin - Toyo International Workshop on Firm Dynamics (Summary)

Chair: Tomohiko Inui (Gakushuin)

Presenter: Davin Chor (Dartmouth)

We investigated whether outward foreign direct investment (FDI) in the banking sector promotes the same country's outward FDI in the non-banking sector. Under the prediction that bank FDI from the same source country decreases FDI costs and promotes manufacturing FDI, we conducted an analysis using Chinese manufacturing firm-level data for 1990 to 2014. We obtained the result that FDI from China's Big 5 banks at t-1 promotes FDI at t by Chinese firms (in manufacturing). Furthermore, even when a variety of sensitivity tests (e.g., changing the analysis timing, estimation using IV) were conducted, the results remained robust.

Discussant: Toshihiro Okubo (Keio)

• The purposes of FDI are various (Greenfield FDI, diversification, etc.)
• Don't credit requirements also vary depending on the purpose?
• Aren't there also other means of financing (issuing bonds, etc.)?
• Isn't there sectoral heterogeneity (differences in dependency on external credit)?
• Don't Chinese government policies have an impact?

Presenter: Yuhei Miyauchi (Stanford/Boston)

Why are economic activities geographically concentrated? Does the firm-to-firm matching rate increase with the number of suppliers due to increasing returns to scale? I conducted an analysis using firm-to-firm data from 2007 to 2016. The results showed that the re-matching rate following unanticipated supplier bankruptcies is higher for firms located in regions of higher agglomeration. I also developed a structural model of firm-to-firm matching, confirming the importance of the mechanism that increases returns to scale.

Discussant: Kenta Yamanouchi (Keio)

• In reduced-form estimation, wouldn't the effects of an earthquake be felt by both buyers and suppliers?
• Shouldn't instrumental variable robustness be checked?
• In structural estimation, isn't it possible that firms with high productivity are likely to relocate to denser cities?

Presenter: Pol Antras (Harvard)

We explain the ‘Spanish export miracle.’ Since the 2008 financial crisis, Spain's GDP has stagnated, but its exports have doubled. We used data on Spanish manufacturing firms from 2002 to 2013 to analyze the relationship between the slump in domestic demand (consumption) and exports. In OLS estimation, a positive correlation was seen between domestic demand and exports, but the results of estimation using instrumental variables showed that the slump in domestic demand led to the increase in exports.

Chair: Shoko Haneda (Chuo)

Presenter: Daisuke Miyakawa (Hitotsubashi)

I used machine learning to analyze firm bankruptcies and exits based on an enormous quantity of firm data. I also analyzed the effect (causal relationship) that variables in firms' supply chain network had on exits. The results showed firm bankruptcies and exits could be predicted very efficiently by using machine learning. Furthermore, In terms of causal relationships, I used double machine learning to confirm the effect that changes in the firm-to-firm networks have on firms' exits.

Discussant: Yuta Takahashi (Hitotsubashi)

• Why did you focus on networks in terms of the effect on firms' exits? There should be many other factors as well.
• What is the decision mechanism for firms' exits?
• Could there be reverse causality?

Presenter: Masaki Hotei (Daito-Bunka)

In Japan, firms with sales at or below 10 million JPY are exempted from paying sales VAT. For that reason, firms report sales at or below 10 million JPY. And for that reason, bunching occurs at the 10 million JPY line. We used TSR data from 2011 to 2014 to analyze what types of firms bunch just below the threshold. The results showed that firms with low productivity and firms with low productivity and high compliance costs bunch just below the threshold.

Discussant: Makoto Hasegawa (Kyoto)

• Could there be a problem with endogeneity?
• Using the median sales as a proxy for productivity, you conclude that a negative coefficient on the median sales implies that low-productivity firms are more likely to bunch. However, couldn't the negative coefficient be more extreme because sales correlate to firm size and thus many firms are affected by the threshold?
• I feel uncomfortable regarding the result that lack of knowledge about the VAT system alone has an effect on bunching.

Chair: Ayako Obashi (Aoyama)

Presenter: In Hwan Jo (NUS)

We focus on intangible assets (organizational capital) as a source of firm growth. What is the role of the accumulation of intangible assets over firm lifecycle? We developed a model of the effect that the presence of intangible assets has on the distribution of resources and aggregate productivity and investigate qualitative impact. The results showed that intangible assets (inter-firm linkages) are correlated to growth and productivity, that they amplify the channel of aggregate productivity shocks, and that sudden deterioration of intangibles triggers recessions.

Discussant: Kaoru Hosono (Gakushuin)

• Can inter-firm linkages be interpreted as organizational capital?
• What are the maintenance costs of organizational capital (linkage)?
• What are the roles of inter-firm linkages in business cycles?
• Why is external finance for investment in physical capital not accounted for in the model?
• Why does TFP increase with age even after controlling for intangibles?

Presenter: Jie Bai (Harvard)

Are there matching frictions in firm-to-firm trade? What is the aggregate loss as a result of these frictions? Could policy interventions facilitate efficient matches? In this study, we conduct an intervention experiment to see whether providing information to suppliers on networks of existing suppliers and buyers leads to higher probability of trade. At present, we are at the stage of preparing the intervention experiment.

Discussant: Kentaro Nakajima (Hitotsubashi)

• Are p and σ identifiable?
• You are saying private information, but can it be bought from TSR? (In other words, isn't this a monetary cost issue?)