Is gold an inflation hedge in Vietnam? A non-linear approach

Figures & data

Figure 1. Gold price (Unit: VND million per troy ounce).

Source: Doji Group.

Figure 2. Bid–ask spread (Unit: VND million per troy ounce).

Source: Doji Group.

Figure 3. Gold price and CPI.

Source: Doji Group, General Statistics Office (GSO).

Note: To compare between gold prices and CPI, all three variables are calculated from the base of 100 as in January of 2011.

Table 1. Unit root test

Variables	Level I(0)			First difference I(1)
	ADF	PP	KPSS	ADF	PP	KPSS
$gpSJ{C}_t$	−0.7485 [0.9669]	−0.7568 [0.9662]	0.3219	−11.9540*** [0.0000]	−11.9588*** [0.0000]	0.1139***
$gp{9999}_t$	−1.2630 [0.8927]	−1.2938 [0.8854]	0.3016	−11.3454*** [0.0000]	−11.3391*** [0.0000]	0.0895***
$cp{i}_t$	−5.7433*** [0.0000]	−7.2349*** [0.0000]	0.2232	−7.0512*** [0.0000]	−6.3426*** [0.0000]	0.2144*

Notes: ***, ** and * are equivalent to 1%, 5% and 10% significance levels. The numbers in brackets [] represent probabilities. gp_t, cpi_t is the natural logarithm of the gold price and the consumer price index, respectively.

Table 3. Structural break

	SJC gold	9999 gold	CPI
No. of breaks	1	1	1
T	03/2020 (0.0000)	01/2020 (0.0000)	10/2016 (0.0000)

Notes: The number of structural breaks is determined by the Bai and Perron (2003) test. T indicates structural break dates. P-values for dummy variables corresponding to identified breaks in the Least Square Regression are in the parenthesis.

Table 5. Estimation of NARDL model (Local gold prices – Global gold price – CPI)

	SJC_CPI&Global		9999_CPI&Global
Variables	$c$	0.3815*** (0.1036)	$c$	1.4598*** (0.1970)
	$gpSJ{C}_{t-1}$	−0.1159*** (0.0303)	$gp{9999}_{t-1}$	−0.4440*** (0.0597)
	$cp{i}_{t-1}^{+}$	0.0849 (0.1041)	$cp{i}_{t-1}^{+}$	−0.0057 (0.0743)
	$cp{i}_{t-1}^{-}$	−0.6152 (0.4047)	$cp{i}_{t-1}^{-}$	−0.3553 (0.2476)
	$gpGloba{l}_{t-1}^{+}$	0.1020*** (0.0304)	$gpGloba{l}_{t-1}^{+}$	0.4552*** (0.0624)
	$gpGloba{l}_{t-1}^{-}$	0.1340*** (0.0310)	$gpGloba{l}_{t-1}^{-}$	0.4683*** (0.0635)
	$\mathrm{\Delta }gpGloba{l}_t^{+}$	0.6029*** (0.0738)	$\mathrm{\Delta }gpGloba{l}_t^{+}$	0.7612*** (0.0534)
	$\mathrm{\Delta }gpGloba{l}_t^{-}$	0.5344*** (0.0889)	$\mathrm{\Delta }gpGloba{l}_t^{-}$	0.7581*** (0.0651)
Long-run coefficients	$L_{gpGlobal}^{+}$	0.8799***	$L_{gpGlobal}^{+}$	1.0367***
	$L_{gpGlobal}^{-}$	1.1560***	$L_{gpGlobal}^{-}$	1.0720***
Criterias	$R^2$	0.6360	$R^2$	0.8343
	Adjusted $R^2$	0.6170	Adjusted $R^2$	0.8257
	AIC (Akaike info criterion)	−4.8669	AIC (Akaike info criterion)	−5.5021
	SIC (Schwarz criterion)	−4.7004	SIC (Schwarz criterion)	−5.3356
	DW (Durbin-Watson stat)	2.0655	DW (Durbin-Watson stat)	1.9744

Notes: For the lagged variables, we only present those with significant coefficients. Standard deviations are in parenthesis. ***, ** and * denote the significance at the 1%, 5% and 10% levels, respectively.

Table 6. Wald test for short- and long-run symmetry (with quarterly data)

	$Long-run{W}_{LR}$	$Short-run{W}_{SR}$	Specification
SJC_CPI	3.9653* [0.0278]	4.1875* [0.0232]	NARDL with LR & SR asymmetry
9999_CPI	10.3397** [0.0003]	2.7168* [0.0796]	NARDL with LR & SR asymmetry

Notes: The associated p-values are in brackets. ***, ** and * indicate the rejection of the null hypotheses of short- and long-run symmetry at the 1%, 5% and 10% levels, respectively.

Table 2. Wald test for short- and long-run symmetry

	$Long-run{W}_{LR}$	$Short-run{W}_{SR}$	Specification
SJC_CPI	5.3453*** [0.0058]	12.3404*** [0.0006]	NARDL with LR & SR asymmetry
9999_CPI	4.2673** [0.0159]	3.5330* [0.0623]	NARDL with LR & SR asymmetry

Notes: The associated p-values are in brackets. ***, ** and * indicate the rejection of the null hypotheses of short- and long-run symmetry at the 1%, 5% and 10% levels, respectively.

Table 7. Estimation of NARDL model (Local gold prices – Global gold price—CPI) (with quarterly data)

	SJC_CPI		9999_CPI
Variables	$c$	0.1966 (0.2266)	$c$	0.9153*** (0.2921)
	$gpSJ{C}_{t-1}$	−0.0882 (0.0603)	$gp{9999}_{t-1}$	−0.2927*** (0.0767)
	$cp{i}_{t-1}^{+}$	0.3078 (0.3212)	$cp{i}_{t-1}^{+}$	−0.1603 (0.4104)
	$cp{i}_{t-1}^{-}$	−0.9259 (3.7319)	$cp{i}_{t-1}^{-}$	−9.0030* (0.4756)
	$\mathrm{\Delta }cp{i}_t^{+}$	2.1855* (1.1383)	$\mathrm{\Delta }cp{i}_t^{+}$	3.0461** (1.3624)
	$\mathrm{\Delta }cp{i}_{t-4}^{+}$	1.5314* (0.8214)	$\mathrm{\Delta }cp{i}_t^{-}$	−11.5056* (5.9225)
	$\mathrm{\Delta }cp{i}_{t-1}^{+}$	−2.1224* (1.1652)	$\mathrm{\Delta }cp{i}_{t-4}^{+}$	2.1430** (0.9471)
Long-run coefficients	$L_{cpi}^{+}$	3.4905	$L_{cpi}^{+}$	−0.5477
	$L_{cpi}^{-}$	−10.5008	$L_{cpi}^{-}$	−30.7582*
Criterias	$R^2$	0.3592	$R^2$	0.3962
	Adjusted $R^2$	0.2523	Adjusted $R^2$	0.2955
	AIC (Akaike info criterion)	−2.9600	AIC (Akaike info criterion)	−2.7072
	SIC (Schwarz criterion)	−2.6733	SIC (Schwarz criterion)	−2.4205
	DW (Durbin-Watson stat)	2.0122	DW (Durbin-Watson stat)	1.9880

Notes: For the lagged variables, we only present those with significant coefficients. Standard deviations are in parenthesis. ***, ** and * denote the significance at the 1%, 5% and 10% levels, respectively.

Table 4. Estimation of NARDL model (Local gold prices and CPI)

	SJC_CPI		9999_CPI
Variables	$c$	0.2949*** (0.1086)	$c$	0.2510*** (0.1050)
	$gpSJ{C}_{t-1}$	−0.0764*** (0.0267)	$gp{9999}_{t-1}$	−0.0680** (0.0265)
	$cp{i}_{t-1}^{+}$	−0.1644* (0.0906)	$cp{i}_{t-1}^{+}$	−0.0929 (0.0859)
	$cp{i}_{t-1}^{-}$	−1.4429** (0.5644)	$cp{i}_{t-1}^{-}$	−0.9458* (0.5062)
	$\mathrm{\Delta }cp{i}_{t-3}^{+}$	1.4911** (0.5822)	$\mathrm{\Delta }cp{i}_t^{+}$	1.2949* (0.6889)
	$\mathrm{\Delta }cp{i}_{t-3}^{-}$	−5.0473** (1.6256)
Long-run coefficients	$L_{cpi}^{+}$	−2.1533*	$L_{cpi}^{+}$	−1.3655
	$L_{cpi}^{-}$	−18.8969**	$L_{cpi}^{-}$	−13.9021*
Criterias	$R^2$	0.1674	$R^2$	0.0752
	Adjusted $R^2$	0.1363	Adjusted $R^2$	0.0482
	AIC (Akaike info criterion)	−4.0576	AIC (Akaike info criterion)	−3.8248
	SIC (Schwarz criterion)	−3.9315	SIC (Schwarz criterion)	−3.7207
	DW (Durbin-Watson stat)	2.0894	DW (Durbin-Watson stat)	1.8341

Notes: For the lagged variables, we only present those with significant coefficients. Standard deviations are in parenthesis. ***, ** and * denote the significance at the 1%, 5% and 10% levels, respectively.

Bai, J., & Perron, P. (2003). Computation and analysis of multiple structural change models. Journal of Applied Econometrics, 18(1), 1–22. https://doi.org/10.1002/jae.659

 Web of Science ®Google Scholar

Data availability statement

The data that support the findings of this study are openly available in figshare at https://doi.org/10.6084/m9.figshare.22720519.v1