Aug 10, 2025

On Chinese Money

If you have ever dealt with Chinese money in the context of banking or trading, you have likely come across various terms such as RMB, CNY, CNH and CNO.

RMB (Renminbi) is an all-inclusive term for all money issued by the People's Bank of China (PBoC). Moving RMB outside of China is possible - but not easy. Therefore, depending on where this money is located, it may have different exchange rate against the USD and also different name. "Location" of money is important.

Part 1: CNY

CNY (onshore) refers to Renminbi held within mainland China. Internal and cross-border transactions involving CNY fall under the jurisdiction of the PBoC.

Foreign companies can hold onshore CNY only if they establish a subsidiary in China, receive payment for goods sold to Chinese customers or participate in certain investment programs.

PBoC enforces strict capital controls and sets a target exchange rate for onshore yuan, because of its effect on trade balance and economic relationships with other countries.

💱All FX transactions executed on the China Foreign Exchange Trade System (CFXS) are subject to market forces (shown as the red line), but cannot deviate more than 2% from the PBoC's policy rate (blue line).

📈The graph below shows that up to August 2023, the PBoC allowed the target rate to float with the market rate on CFXS - signaling no active intervention.

☝However, since August 2023, the PBoC began setting the target rate below the CFXS market average, indicating a "strengthening policy" for the yuan.

🗣️President Trump accused China of deliberately devaluing the yuan to make exports cheaper - specifically in August 2019 and again in April 2025. This claim may be broadly based on existence of capital controls, however is not supported by the below graph or interest rate data, which speaks opposite.


Technically speaking, no currency ever physically leaves the country where it's issued - and China is no exception. Foreign banks hold RMB in nostro accounts within mainland institutions. So, when we refer to "inside" or "outside" China, we're speaking in regulatory terms, not geographical ones.

Part 2: CNH

Before the introduction of CNH, the only way to move money in and out of China was to convert it into USD or another currency using the onshore China Foreign Exchange Trade System (CFXS), under PBoC supervision. Trading onshore yuan is subject to capital controls, and it is not possible to freely send CNY directly between foreign banks.

🌐 As part of the internationalisation effort, a "free version" of yuan was introduced in 2010. Crossing the onshore-offshore boundary is still subject to the same capital controls, but once the currency is border-cleared, it can be traded much like any other G10 currency.

The offshore version of yuan is called CNH, where "H" stands for Hong Kong - the first place where CNY was originally internationalised. Nowadays, banks in many other international centres accept CNH deposits.

1 CNY becomes 1 CNH once it crosses the onshore-offshore boundary. However, because crossing this boundary is not easy, yuan can have different values in different geographical locations. The graph below shows USD/CNY and USD/CNH rates.

The graph illustrates several points:

1️⃣ Overnight CNH-CNY premium fluctuations are very common. One day, CNH trades at a 1.3% discount to CNY; the next day, the situation reverses. This is because CNH is market-driven and reacts faster to international news, whereas CNY is managed by the PBoC and trades under supervision.

2️⃣ A 90-day rolling mean shows that the CNY/CNH premium tends to exhibit seasonality. We can conclude that, since 2023, onshore yuan tends to be, on average, more valuable than its offshore counterpart. This indicates that crossing the offshore-to-onshore boundary inward may be marginally more difficult than the other way around.

3️⃣ The green line represents the premium that 3-month CNH deposits in Hong Kong banks command over CNY deposits in Shanghai banks (i.e., HIBOR minus SHIBOR). As shown, CNH deposits generally offer higher interest rates.

Furthermore, CNH deposit premiums tend to be negatively correlated with CNH spot FX premiums. This relationship can be explained by interest rate parity, as both currencies are ultimately expected to converge to a one-to-one exchange rate.

4️⃣ Last but not least, I believe that arbitraging any sort of disparity between CNY and CNH is next to impossible. The observed behaviour is, to a large extent, a “real money” matter.


Part-3: Non-deliverables

CNY refers to Renminbi deposits held in mainland Chinese banks, while CNH represents similar deposits in Hong Kong and other offshore financial institutions. Although the nominal value of CNY and CNH is 1:1, capital controls can cause exchange rates to diverge across jurisdictions.

What do CNY and CNH have in common? Both represent real money that can be physically delivered into accounts and earn interest—SHIBOR in the case of Shanghai banks, and HIBOR for Hong Kong banks.

🌍The world of non-deliverables🌍

In today’s post, we’ll explore the non-deliverable yuan and its associated derivatives market. In some banks, this is denoted by the currency code CNO. CNO differs from both CNY and CNH. It cannot be delivered or deposited anywhere. There is no spot market for it, nor any interbank interest rate.

How non-deliverability work ?

In a non-deliverable forward (NDF), two parties - say, located in London -agree on a forward FX transaction for a specific amount to be settled on a future date. However, since neither party can access actual yuan, the contract is typically settled in USD, based on the difference between the agreed forward rate and the prevailing spot rate at settlement.

The attached chart (past 2 years) shows several important points:

1️⃣ Both CNY and CNH (🟦+🟥) show similar patterns when it comes to FX forward premiums. The premiums are typically negative, due to the fact that USD is generally a high-yielding currency. The graph also shows a small cross-currency basis, compared to what we would get under the HIBOR vs. SHIBOR interest rate differential.

2️⃣ SHIBOR and HIBOR rates are strong "real money" factors keeping FX forward premia in check. These interbank rates help anchor expectations and reduce volatility in deliverable forward markets. In contrast, the non-deliverable forward premia (CNO 🟩) lack such stabilizing influences, making them considerably more erratic

3️⃣ While deliverable CNY forwards (🟥) trade as spread to USD/CNY CFXS spots (as one would naturally expect), the non-deliverable forwards are settled against the PBoC’s policy FX rate instead (🟩).

☝This has broad implications for the non-deliverable market. By now, you should know the PBoC’s policy FX rate is not fully market-driven and can occasionally be influenced by policymakers’ discretion - an influence that also extends to non-deliverable forward payoffs.

🤔 Why is that the case? Is it a legacy of non-deliverable market? Or does this convention has a reason? One thing is certain: the market has it's own mind, as it shows a stronger correlation between outright NDF rates and the CFXS spot rate (92.8%), than with the PBoC fixing rate (81.8%).



May 21, 2025

A word on "cross-currency basis"

A few weeks back, ESTER/SOFR cross-currency basis briefly crossed zero before retreating back into negative territory. This event flashed in news, but details were omitted. I will try to fill-in the gaps, using the latest market data: 

Have you ever wondered ?

1) What does cross-currency basis mean? 
2) Why it exists? 
3) Why it has been traditionally negative?

When I did my CFA more than 10 years ago, this basis was not mentioned anywhere in the curriculum. Instead, all three levels were grinding covered interest rate parity (IRP). According to this parity, the basis shouldn't exist:
FxFwd / FxSpot = 1 + (r1-r2) × n/360
Later, quants gave me the following explanation: 
🗣️ "Cross-currency discount curve (such as EUR) is used to discount EUR leg of an xccy swap, collateralized using USD rate. We fit this curve in order to correctly reprice both legs of xccy swap". 
Next, traders on desk told me: 
🗣️ "The basis is caused by supply and demand" 
I found none of these explanations helpful because they don't carry any economic meaning, nor do they reconcile with what I learned about IRP earlier. 

So, here is the main reason why this basis exists and IRP is seemingly broken: 

Funding preferences on credit markets


The covered interest rate parity (IRP) condition holds in simplified, risk-free settings (e.g., SOFR vs. ESTER). However, in real-world markets, most cross-currency transactions occur on the credit side beyond the reach of any parity/arbitrage rules. 

 If you are German AAA-rated corporation seeking 5-year funding as of 20/5/2025, you have two options: 

1) Borrow in EUR at the risk-free rate of 2.14% plus a 46 bp credit spread (all-in 2.60%). 

2) Borrow in USD at the risk-free rate of 4.06% plus a 29 bp credit spread (all-in 4.35%) + hedge USD exposure using 5Y FxSwap 

Without credit spreads, the risk-free rate differential (4.06% - 2.14% = 1.92%) would be offset by FX fwd premium (remember IRP), leaving the issuer indifferent between markets.

However, the USD AAA-rated market has historically offered a lower credit premium (here, 17 bp less than EUR), making it more attractive for international borrowers. This demand dislocated interest rate parity, requiring EUR/USD cross-currency market participants to give-up 8 bp on their ESTER rate to execute the trade. 

The attached graph shows that cross-currency basis on derivatives market has been traditionally correlated with the differential of credit risk premium in respective countries. Yet because credit markets are incomplete, this relationship holds only in a weaker form - no strict arbitrage forces these rates to converge perfectly. 

But how about the arbitrage ? 


Executing the IRP arbitrage effectively requires significant leverage. 

XCY swaps are essentially the only instruments which can be used to effectively arbitrage the long-end of the yield curve. However here is the catch: they are all collateralised in USD regardless of which counterparty is posting collateral. 

In fact, the asymmetric demand for USD collateral is therefore another reason why cross-currency basis exists. 

That closes the circle.

Jun 24, 2020

Factor Investing and Fama-French model

This notebook illustrates factor investing and five-factor Fama-French model.


Risk Factor

Certain characteristic of economy (Inflation/GDP) or stock market itself (S&P 500)

Factor Model

Factor model uses movements in risk factors to explains portfolio returns

Questions which factor investing answers

  • Why different asset have systematically lower or higher average returns?
  • How to manage the asset portfolio with the underlying risks in mind?
  • How to benefit of our ability to bear specific types of risks to generate returns?

Fama-French Model


Assumes linear relationship between empirical factors and stock returns:

  • Market Factor (MER)
  • Size Factor (SMB)
  • Value Factor (HML)
  • Profitability Factor (RMW)
  • Investment Factor (CMA)

Factors are constructed daily from definitions, as illustrated previously

  • They are global for the entire stock market

Factor sensitivities are calibrated using regression

  • They represent “reward for taking a specific risk”, which is different for every stock
  • Risk/Reward relationship is expected to hold over time
  • Objective: maximize the model’s predictive power R2

Market Excess Return (MER)

  • Market excess return (over RF rate) alone explains around 80% of asset movements
  • Daily returns are ~normally distributed
  • Relationship between returns of the overall market and returns of selected portfolio

Size (SMB) factor

  • Small-cap companies typically bear additional risk premium - was it always the case?
  • Python can help you to see that this factor has a different prevalence in different economic regimes

Value (HML) factor

  • Value companies trade at higher yields to compensate for lack of growth potential
  • Python can help you to see that this factor has different explanatory power in different market situations and on different portfolios (very interesting)

Profitability and investment factors

  • Profitability factor (RMW) to attribute superior returns of companies with robust operating profit margins and strong competitive position among peers

  • Investment factor (CMA) to segment companies based on their capital expenditures

  • Analysts opinion: High capex structurally associated with growth companies, which puts usefulness of this factor in question


Evaluating 5-factor model

  • Analyst opinion: High correlations between risk factors puts usefulness of 5-factor model into question.
  • R2 10-20% for RMW, CMA
  • 5 factor improvement only by 0.2%