International Workshop

Uncertainty, Trade and Firms (Summary)


    • Time and Date: 9:50-18:00, April 18, 2018
    • Venue: RIETI's seminar room #1121 (METI Annex 11th Floor)
    • Language: English
    • Host(s): Research Institute of Economy, Trade and Industry (RIETI) / JSPS KAKENHI (B) (Grant Number: 17H02531)


"Business Level Expectations and Uncertainty"

Nicholas BLOOM (Stanford University)

Presentation of cutting-edge research results regarding expectation formation, future predictions, and uncertainty by American manufacturing plants, based on the Management and Organizational Practices Survey (MOPS) carried out as part of the U.S. census in 2015. The underlying survey was an attempt to measure the uncertainty faced by plants by asking them about sales, capital investment, employment, materials and other costs, and about the subjective probability of five future scenarios, and the presentation is a cross-section of business level data analysis regarding the validity of the survey's methods, the robustness of its results, as well as measurement errors. In particular, the positive correlation between recent sales growth rates and future sales expectations and the positive correlation between past sales volatility and the distribution of future sales expectations were pointed out, followed by debate on the validity of the methods employed by the survey and its usefulness.

A lively Q&A and discussion included questions from the floor regarding the occupation of the respondents questioned in the survey and about the applicability of the survey to the service sector, a large part of the economy.

"Policy Uncertainty in Japan"

Steven DAVIS (University of Chicago)

Uncertainty regarding policy, based on factors such as the repeated delay in raising the consumption tax rate and the inauguration of the Trump administration in the United States, has garnered much interest. Steven Davis reports on research that measures policy uncertainty in Japan based on newspaper reporting. Davis compiled a policy uncertainty index by using articles published in four major newspapers, such as the Nikkei and Yomiuri Shimbun, specifically articles that mention the "economy," "economic environment", "opacity", "uncertainty", "anxiety" and "concern," and containing vocabulary pertaining to policy (e.g. "government debt," "Bank of Japan," and "regulations"). The three following points have primarily become obvious from the index: First, when comparing this index with the political instability index created using the approval ratings of political parties from opinion surveys, one notices the trend that policy uncertainty is low when politics is stable. Second, the majority of policy uncertainty arises from matters pertaining to fiscal and monetary policy. Third, when estimating a VAR model that includes the policy uncertainty index, macroeconomic performance worsens when a positive uncertainty shock occurs. However, there are multiple interpretations of this result.

After the presentation there was a question regarding how appropriate it is to use newspaper articles to quantify policy uncertainty and a proposal to include exchange rates in the VAR model. There was lively debate regarding the research.

"The Effects of Firm Uncertainty on Economic Activity: New Evidence from One Million Documents"

Kyle HANDLEY (University of Michigan)

Companies face various uncertainties (policy, macro-level risks, firm-specific risks), which influence their investment and employment, but measuring firm-level uncertainty is difficult. The present research has constructed a firm-level uncertainty index using business documents of approximately 40,000 firms obligated to report to the United States' Securities and Exchange Commission (SEC) between 1994 and 2016. The index is the rate of terms like uncertainty and ambiguity as a portion of the overall number of terms in the documents. The researcher, Kyle Handley, then conducted an empirical analysis of the relationship of this rate to investment activity of these firms. Based on the results of this analysis, a firm's investment rate drops by approximately 0.5 percentage points for each standard deviation increase in the firm-level uncertainty index. It also became apparent that uncertainty, aggregated chronologically, negatively correlates to macro-level variables (investment, GDP, employment etc.).

After the presentation there were comments such as whether the regression analysis should include firms' fixed effects.

"Uncertainty over Working Schedules and Compensating Wage Differentials: From the Viewpoint of Labor Management"


Much research has been done on long work hours, but there is little research dealing with the uncertainty of work hours due to unexpected overtime and the like. Masayuki Morikawa's research uses original individual survey data to present observational evidence regarding the uncertainty of work schedules in Japan. According to Morikawa's results, about 50% of workers have experienced unanticipated overtime work, while about 30% are occasionally forced to cancel scheduled holidays due to sudden work issues. Such uncertainty in work schedules is prevalent among fulltime regular employees and those working long hours. For workers, the cost of this uncertainty is large, with unscheduled overtime equivalent to 150% or more of scheduled overtime amount. In addition, the negative effect of the uncertainty on work satisfaction is much greater than the negative effects of increased working hours or decreases in wages is extremely large. Although the existence of wage premium compensation for the uncertainty of overtime can be observed in the real labor market, its quantity is small.

The presentation was followed by productive comments for future improvement of the survey, including the possibility of utilizing more objective terminology, for instance "how many times a year," in place of the subjective terminology like whether uncertain labor happens "frequently" or "occasionally," that the survey should consider the relationship to the glass ceiling for women, or that the survey could be combined with one aimed at companies.

"Using Newspaper Text to Quantify Geopolitical Uncertainty"


Among factors such as the inauguration of the Trump administration in the United States or rising military tensions between the United States and North Korea, geopolitical risks have recently garnered attention. However, there is no index to measure such risks. Because of this, policymakers are unable to quantitatively evaluate geopolitical risk. This presentation reports on research quantifying the opacity of future economic developments arising from political or military events and tensions happening in the world, based on newspaper reporting. A newly created index (called "an index of geopolitical-related economic uncertainty" in the research paper) is based on articles published in four major newspapers, such as the Nikkei and Yomiuri Shimbun, especially articles containing terms relating to the "economy" or "economic environment," "anxiety," "uncertainty," or "concern," as well as vocabulary directly indicating political and military events and rising tensions (e. g. "air strike," "growing tension," and "presidential election"). Based on this index, it can first be said that the index rose substantially during the gulf war, the Iraq war, the 2016 U.S. presidential election, and the 2017 U.S. airstrikes against Syria as well as the rising tensions between the U.S. and North Korea, and that, second, based on the estimations of a VAR model including the index, macroeconomic performance falls when positive uncertainty shocks occur.

After the presentation it was pointed out that it was unclear what the index seeks to capture and that the concept should be made clearer. There were also proposals to randomly sample articles containing vocabulary related to the two categories of economics and unpredictability and to create a list of base terminology that indicate political and military events and rising tensions based on the results of carefully studying such articles, and to include exchange rates in the VAR model.

"Business Plans and Expectations Survey: First Results"

SENGA Tatsuro (RIETI / Queen Mary University of London)

Analysis results of the Business Plans and Expectations Survey, conducted at RIETI in 2017, were presented. The survey was supported by KAKENHI (Grant-in-Aid for Scientific Research) and attempted to measure the uncertainty faced by Japanese businesses, utilizing the Management and Organizational Practices Survey (MOPS), whose results were introduced during Nicholas Bloom's presentation, as a reference. Like MOPS, this survey compiled a subjective probability distribution of answers regarding five future scenarios and their respective probability of occurrence according to the survey subjects. As a new approach, questions using the Likert scale to measure uncertainty faced by firms and questions about firms' positive or negative outlooks were added to the survey. Interim analysis results that were observed included the fact that uncertainty, as measured by the Likert scale, has a negative impact on capital investment and hiring plans of businesses, and that even when prediction center values are level, businesses that indicated more pessimistic predictions of the future had lower hiring plans than ones that indicated more optimistic future predictions.

Audience members pointed out that the possibility of measuring uncertainty utilizing the Likert scale was beneficial for future survey design in light of its simplicity. It was also pointed out that additional examination of the findings and methodology is necessary due to the fact that in terms of the survey items that were used to measure the optimism or pessimism of the respondents, there was the possibility of a discrepancy between what the survey is really attempting to measure and what the responding firms actually meant through their answers.

"Uncertainty, Imperfect Information and Learning in the International Market"

Cheng CHEN (University of Hong Kong)

Among circumstances such as the Trump administration's U.S. trade policy and Britain's withdrawal from the European Union, uncertainty regarding the future of the global economy is rising. However, there is little research on the impact of such uncertainty affecting foreign markets on the exports and foreign direct investment by Japanese companies, and policy proposals founded on empirical analyses based on data are equally rare. The present research uses data on Japanese multinational corporations that includes information on sales predictions for foreign subsidiaries (the Ministry of Economy, Trade and Industry's "Basic Survey on Overseas Business Activities" and "Basic Survey of Japanese Business Structure and Activities") to offer four new facts regarding the uncertainty faced by companies in foreign markets. First, there is a positive correlation between uncertainty faced by foreign subsidiaries and macro-level uncertainty in the countries in which they are located. Second, foreign subsidiaries predict sales more accurately the more years they have been operating. Third, when first expanding to a foreign country through direct investment, compared to foreign subsidiaries of companies with no experience in exporting to the region (Asia, Europe, North America, etc.) that the country is located in, foreign subsidiaries of companies that do have experience in exporting to that region are able to more accurately predict local demand. Fourth, there is a positive correlation between prediction errors for the present term and for the following term and this correlation increases proportionally to the distance from Japan. These discoveries are direct evidence that companies learn about foreign demand uncertainty through subsidiaries' sales and exports. Furthermore, starting with a dynamic model of corporate growth, the present research expanded this model to have companies provide goods to foreign markets by choosing exports or direct investment. The calibrated model not only incorporates the four above-mentioned observed facts, but can also reproduce characteristics of the dynamics of exports and direct investment. In addition, analysis results of counterfactual experiments imply that when analyzing the impact of changes in foreign demand uncertainty or trade liberalization on international trade and the activities of multinational corporations, it is important to consider companies' predictions and learning.

One comment mentioned that because maritime data contain various industries (all manufacturing industries and some service industries), it would be interesting to examine how the above-mentioned observed facts change based on the industry in question. There was also a question regarding how the average and distribution of companies' prediction errors changes based on the time period (e.g. during the global financial crisis and the Southeast Asian financial crisis). The presenter plans to make this question a research topic in a future project.

"Factor-Biased Multinational Production and the Labor Share"

Chang SUN (University of Hong Kong)

The present research uses a dataset of global companies (i.e. multinational corporations) to provide two new observed facts regarding capital intensity and wage structures of multinational corporations. First, multinational corporations are firms of a large scale and employ capital-intensive technology. Compared with non-multinational corporations, multinational corporations are more capital-intensive and have a lower wage-capital ratio. Second, multinational corporations with parent companies in capital-rich countries use more capital-intensive technology than multinational corporations where this is not the case, and transfer more of this technology from the parent company to subsidiaries. Furthermore, the present research expands the traditional multinational corporation model to include in the model the possibility that multinational corporations choose technology of different capital intensity in different countries. The research also discusses how the liberalization of direct investment affects the global wage distribution (the ratio of wages to capital). The calibrated model reproduces not only the two above-mentioned observed facts, but also patterns of decreases in the global wage-capital ratio. Analysis results from counterfactual experiments suggest that expansions of activities by multinational corporations lead to a falling global wage-capital ratio.

Responding to a question from the floor regarding whether in the model multinational corporations' capital can move across national borders, Chang Sun explained a case in which he considered both possibilities that capital could and could not move across national borders. Someone also pointed out that there is a strong chance that subsidiaries of multinational corporations may have lifecycle dynamics and with time use more local capital. Sun responded that he should incorporate this point into his model for a future paper.

"Production Chains, Exchange Rate Shocks and Firm Performance"

ZHANG Hongyong (RIETI)

Though large fluctuations in exchange rates not only substantially impact exporters' performance, but through ripple effects can also cause macroeconomic fluctuations, there is extremely little research dealing with the ripple effects of exchange rate fluctuations. The present research uses data on domestic production networks of Japanese companies and data on international trade to investigate ripple effects of exchange rate fluctuations on production and companies upstream and downstream via supply chain networks between companies. As an exchange rate shock, the research used corporate level import and export data to calculate individual companies' effective exchange rate exposure. Based on results of empirical analyses, it found that the statistically significant effect of importers' exchange rate exposures on the sales or profit ratios of their buyers (indirect importers which do not themselves import) is small, but that exporters' exchange rate exposures had an extremely large impact on the sales and profit ratios of suppliers (indirect exporters which do not themselves export). When the yen is weak, the sales and profit ratios of indirect importers do not fall significantly, but the sales and profit ratios of indirect exporters increase substantially. In particular, small and medium-sized enterprises supplying goods to exporters are easily affected by large exporting companies' exchange rate exposure. Our research suggests that, viewed from the perspective of the domestic supply chain, a stabilization of exchange rates is extremely important for the performance of Japanese companies, in particular for small and medium-sized enterprises engaging in indirect export.

After the presentation, there were many fruitful comments, such as regarding the definition of ripple effects upstream and downstream and the interpretation of results, the creation of exchange rate exposure variables and export amounts, and whether there aren't also companies engaging in parallel importation.