“It May Be the Biggest Tax Heist Ever.”, wrote the New York Times on January, 20th 2020. “It” being so-called cum-ex trades in which traders file for dividend tax refunds for taxes that they never paid. Surprisingly, several European Treasuries did (and do) indeed refund these never-paid-taxes. Investigations by a consortium of journalists estimated these refunds at around $60 billion per year and argue that this tax fraud (in some variations) is still ongoing in 2021.
Treasuries seemed to be confused about who paid the tax and who is eligible for a refund, largely because the tax was withheld by the corporation issuing the shares but the certificate for tax reimbursement was issued by custodian banks (e.g., in Germany till 2012).
Investors exploited this confusion and tax-loop hole by colluding to artificially create two legal owners of the same share certificate before the ex-dividend date, i.e., before the date on which the share is not eligible for the dividend payment anymore, assuming trading with standard settlement conditions. To receive a dividend an investor must hold the share on the record date, considering that shares often settle in two days, the ex-dividend date is often two days before the record date. These exploits are called cum-ex trades.
What are cum-ex trades?
As we elaborate in more detail in our recent paper, cum-ex trades are often done by three colluding investors: Investor A holds shares of firm X and on the record date (T) will receive dividends less mandatory withholding taxes (WHT), and a generic tax certificate for these WHT, missing any details linked to the specific transaction. Before the ex-date, on date T-3, Investor B short-sells shares of firm X cum-dividend to investor C to be delivered on the registration date T. On the registration date investor B buys the shares from investor A with same-day special settlement conditions, which investor B can then deliver in time, i.e., on date T, to investor C. Because these shares are ex-dividend, Investor B also provides cash compensation for the dividends less mandatory WHT. On date T+1 investor C can sell the shares back to investor A.
Because, according to German (and potentially other countries) tax laws, investor C is the “economic owner” on the registration date, investor C also receives a tax certificate, which allows investor C to claim back WHT that were never paid. Investor B did not pay dividends and did not pay taxes to the government, Investor B merely paid Investor C a cash compensation for the forgone dividend payment. In 2007, the Annual Tax Law in Germany closed this loophole, but still allowed to these trades to run through custodial banks abroad (until 2012).
Our question: Did investors use ADRs for cum-ex trades?
We investigate whether American Depositary Receipts (ADRs) were used for cum-ex deals from 1999 to 2007. In particular, we investigate trading by institutions, which allows us to identify changes in portfolio holdings and therefore better allows us to pinpoint potential trading motives.
ADRs seem to be particular suitable for this kind of tax fraud because they are a negotiable receipt for a foreign (often European) security which can be traded in the US the same way as common shares. The ability of investors to create new ADRs out of thin air decreases short-selling frictions which should ease cum-ex trading.
Since 2014, the US Securities and Exchange Commission (SEC) started investigating the pre-release of ADRs and found “industry-wide abuses,” fining various firms more than half a billion dollars. Pre-released ADRs are released before the foreign security was deposited with the deposit agent. While legal, the pre-release of ADRs is strictly regulated. In particular, either the pre-release agent or its client must hold the foreign securities to avoid inflating the total number of ADRs and foreign shares available for trading. As elaborated by the SEC, the pre-release of ADRs “inflated the total number of a foreign issuer’s tradeable securities and resulted in abusive practices such as inappropriate short selling and dividend arbitrage. In certain countries, demand for ADR borrowing increased around dividend record dates, so that certain tax-advantaged borrowers could – through a series of transactions – collect dividends without any corresponding tax withholding.”
The evidence.
Around ex-dividend dates, we document a market-wide ADR abnormal turnover across all US exchanges of around 17%. We calculate abnormal turnover as the average turnover (shares traded over shares outstanding) during the event period (5-days before to 5-days after the event) dividend by the average turnover in the benchmark period (45-days before to 45-days after the event, but excluding the event period) minus one. We further find that:
- Turnover from trades with special settlement conditions increases by around 14%.
- Turnover from institutions increases by around 166%.
The substantial increase in institutional trading could be motivated to capture or avoid the dividend. Because of potential differential tax treatments of dividends and dividend substitutes, certain investors might prefer one to the other. While we find some evidence for both trading strategies, several findings are inconsistent and cannot be explained by either dividend capture or dividend avoidance motivated trading.
Several findings are consistent with explaining this increase in ADR trading volume by cum-ex deals. Around 600 institutions trade ADRs around ex-dividend dates, of which less than half day trade, i.e., buy and sell the same ADR on the same day. Day trading is often a requirement for both dividend capture or dividend avoidance-motivated trading. The majority of institutions does not day trade and therefore seems to have other motives for trading around ex-dividend dates.
The average institution sells before the ex-dividend date and buys after the ex-dividend date. The average institution sells ADRs worth around $100 million three days, two days, and one day before the ex-dividend date. On the ex-dividend date and the day afterwards, institutions buy ADRs worth around $USD 400 million, on average.
ADR trading volume from trades with special settlement conditions is abnormally high on the ex-dividend date and the day afterwards, suggesting that institutions use special settlement conditions. As with cum-ex deals, institutions seem to (short) sell ADRs before the ex-dividend and buy them back with special settlement conditions afterwards.
Conclusion.
While our evidence for the motivation behind why institutions trade around ex-dividend dates is far from conclusive, the significant increase in institutional trading volume; the industry wide abuse of ADR pre-releases; and the billions of tax dollars lost to several, mainly European, Treasuries, is a fact and requires further investigation.
Jonathan Brogaard is a Professor at the University of Utah David Eccles School of Business
Dominik Roesch is an Assistant Professor at the University of Buffalo School of Management
This post is adapted from their paper, “Tax heist using American Depositary Receipts” available on SSRN.
The views expressed in this post are those of the authors and do not represent the views of the Global Financial Markets Center or Duke Law.
It’s truly amazing at the lengths that individuals and companies will go through for money. These actions aren’t illegal, but there is no question it goes against the spirit of the law and needs to be visited. The amount in question is also quite staggering.