A Novel Methodology for Credit Spread Prediction: Depth-Gated Recurrent Neural Network with Self-Attention Mechanism

Table 2

Variables and data.

Variables		Data
Credit spread		ICE BofAML US corporate master option-adjusted spread^$
Financial market factors	Credit spread term structure	Level factor	The first principle component extracted
			From a large cross section of credit spreads of various maturities and various ratings^$
		Slop factor	The second principal component extracted
			From a large cross section of credit spreads of various maturities^$
		Curve factor	The third principle component extracted
			From a large cross section of credit spreads of various maturities^$
	Risk-free interest rate term structure	Level factor	The long-term factor of Nelson and Siegel term structure decomposition extracted from a large cross section of treasury yields of various maturities^$$
		Slop factor	The short-term factor of Nelson and Siegel term structure decomposition extracted from a large cross section of treasury yields of various maturities^$$
		Curve factor	The mid-term factor of Nelson and Siegel term structure decomposition extracted from a large cross section of treasury yields of various maturities^$$
	Fama-French factor returns	Excess return on the market
		Small-minus-big return
		High-minus-low return
	Return on stock index	Return on S&P 500
	Volatility of stock index	Volatility of S&P 500
	Exchange rate	U.S. Dollar index
	Oil prices	Crude oil prices: West Texas intermediate (WTI)
	TED spread	Difference between 3-month LIBOR based on U.S. dollars and 3-month treasury bill
	Swap spread	Difference between10-year swap rate and 10-year treasury yield
	Commodity price index	RJ/CRB index

Note: Data are obtained from https://fred.stlouisfed.org; $ ICE BofAML US Corporate 1–3\3–5\5–7\7–10\10–15\15+ \AAA\AA\A\BBB\BB\B\CCC Option-Adjusted Spread;$$ 3\6-month and 1\2\3\5\10-year treasury yields.