Research

Publications

"The U.S. Treasury Premium," with Jesse Schreger and Wenxin Du. Journal of International Economics, 2018 (Link)

• • Peer-reviewed in top Q1 journal

  • 350+ citations

  • Media Coverage: NBER Digest, VoxEu

We quantify the difference in the convenience yield of U.S. Treasuries and government bonds of other developed countries by measuring the deviation from covered interest parity between government bond yields. We call this wedge the “U.S. Treasury Premium.” We document a secular decline in the U.S. Treasury Premium at medium to long maturities. The five-year U.S. Treasury Premium averages approximately 21 basis points prior to the Global Financial Crisis, increases up to 90 basis points during the crisis, and has disappeared after the crisis with the post-crisis mean at-8 basis points. Meanwhile, the short-term U.S. Treasury Premium remains positive post-crisis. We discuss the impact of sovereign credit risk, FX swap market frictions, and the relative supply of government bonds on the U.S. Treasury Premium. 

Files:

• Stata Dataset, Excel Dataset

• Excel Data Supplement: list of Bloomberg and Datastream tickers with instructions for reproducing the data file. 

• Data Appendix

• NBER IFM Data Session Slides 

If you use this data, please cite:

Du, Wenxin and Jesse Schreger (2016). “Local Currency Sovereign Risk.” Journal of Finance, 71, 1027-1070.

Du, Wenxin, Joanne Im, and Jesse Schreger (2018). “The U.S. Treasury Premium.” Journal of International Economics, 112, 167-181.

Pre-print

International Finance

• “Real Bond Parity" (Link)
  
  Summary: I test a set of assumptions that imply the return parity of long-run, real bonds denominated in different currency numeraire. The joint hypothesis is rejected in our post-2009 sample of developing and developed market currencies; however, I document a strong relationship between changes in the log of bilateral, real exchange rates and real holding period bond returns in the direction of parity, contributing to the Meese-Rogoff puzzle on exchange rate determination.

• "The Climate and Financial Effects of Fossil Fuel Power Plant Sales in the U.S."  (Responding to feedback, available on request)

Summary:  Recent pressure on publicly traded firms to decarbonize and their decisions to high carbon, “dirty” assets to private firms have sparked concerns about the climate and financial consequences of these sales. Using a dataset of fossil fuel power plant M&A deals between 2000-2022, I estimate sale effects on production and emissions and announcement effects on seller valuations. I find that public to private sales had near zero effects on asset emissions and that these effects were statistically indistinguishable from effects in public to public sales. I then present a simple model of firm production and emissions in which publicly traded firms, but not privately held firms, experience a shock to their private costs of emitting. The results suggest that ESG pressure that lead to firm level emissions reductions via asset sales may have near-zero climate impacts.

Works-in-progress

Machine Learning/NLP

• "Distribution Shifts in Review Classification by BERT and T5 (NLP Models)," with Sehaj Chawla and Eduardo Boratto, 2021. (Link) (Received co-authorship invitation from course instructor of graduate MIT machine learning class).

Summary: We test the generalizability of algorithms that detect fake reviews based on review text by exploring the effects of distribution shifts (with respect to time, industry type, product type, and sentiment) on the performance of four pre-trained and fine-tuned transformer models. The first three models are neural nets built on top of three pre-trained instances of BERT (large, small, and mobile), which generate contextualised embedding of review text in a way that is mechanically independent from our training dataset. Our fourth model is the small T5 transformer.

Climate Finance

• "Evaluating Schemes to Green Battery Arbitrage," with Thomas Lee. 

Summary: We derive conditions under which a popular policy for greening electricity battery arbitrage–compensating batteries for the marginal emissions (ME) they take off the grid–increases, decreases, or has no effect on aggregate emissions from electricity generation.

Mini Side Projects

• Case study on effects of climate change on the cyclonic activity and thus, precipitation in the Mediterranean and water flow in the Tigris and Euphrates Rivers--the primary sources of potable water in the Mesopotamian region.  Slides.

  • Researched and written with Minghao Chen for a master's level course on climate change in the MIT Civil and Environmental Engineering Department taught by Professor Elfatih Eltahir, a key contributor to the MIT Regional Climate Models.  

• Estimate of the green premium in the U.S. dollar, euro, and Hong Kong dollar in 2021. Link.

  • Note: In April 2023, "Germany sold one of the year’s biggest green bonds in April, a €5.25bn 10-year security. As with other green German sovereign bonds, the security is twinned with a non-green transaction that has a matching maturity. This allows for a simple comparison in pricing between the two – what is sometimes known as the ‘greenium’ – and in this case the new green bond priced just half a basis point inside its twin."  Link. 

• Dataset of the parent ownership of and annual emissions high greenhouse gas emitting facilities in the U.S. from 2011-2022 at the facility level, and parent characteristics. 

  • Required extensive cleaning of EPA's GHGRP ownership dataset, and matching with time-dependent records of parent-subsidiary relations from CorpWatch, OpenCorps,  Google, Wikipedia. 

  • Shows high match with aggregated result published by U. Mass Amhert PERI's Greenhouse 100 Index.