Authors:

Taabeenah Riyaz, Garima Tak, Parth Agrawal, Shambhavi, Aman Arya, Sanjyot Rane

ABSTRACT

The Iran–Israel conflict window (1 December 2025-15 May 2026) used in this study has been revisited to see that the safe haven characteristics of gold continued to hold under current financial market conditions. Gold is well established as the preferred asset during geopolitical crisis, but a variety of studies – ranging from Bauer and Lucey’s (2010) formal distinction between hedge and safe-haven asset to Madhavan’s (2023) findings of fading Indian safe-haven effect – indicate that this safe-haven status is not automatic or permanent across crises. The present research examines this uncertainty in a conflict never studied before; even though the geopolitical situation remained fairly stable, gold did not show a monotonic appreciation path but rather a sharp initial rise with two easily seen short-term corrections, which leads to the question of whether the classical safe-haven theory is sufficient to explain gold’s behavior. The study utilizes 166 daily observations of gold, Brent crude oil, US Dollar Index, Federal Funds Rate, US CPI inflation and 10 year Treasury yields, and an econometric design that consists of five theoretical perspectives: Safe-Haven Theory, Portfolio Theory, Flight-to-Quality Theory, the Efficient Market Hypothesis and Behavioural Finance, along with descriptive statistics and Augmented Dickey-Fuller (ADF) unit-root testing, multivariate Ordinary Least Squares (OLS) regression and Multivariate GARCH (MGARCH) covariance modelling. The results show that gold’s price level was significantly and negatively associated with the US Dollar Index (β = −125.94, p < 0.001), the Federal Funds Rate (β = −5,907.13, p < 0.001) and 10-year Treasury yields (β = −1,630.40, p < 0.001), while Brent crude oil carried a significant positive coefficient (β = 9.20, p < 0.001); inflation was statistically insignificant (p = 0.303). The model accounted for 71.1 percent of the variation in gold’s price (R² = 0.711). In terms of geopolitical risk pricing, opportunity-cost and liquidity-preference channels continued to dominate gold’s short-term movements, as suggested by the fact that the estimates of the MGARCH model show that gold was negatively correlated with Treasuries. The results suggest that the safe-haven effect of gold during the conflict in 2026 was not an unconditional one, but only a return effect; hence, it was a geopolitical hedge in terms of returns but was structurally open to monetary-policy and currency effects. This has implications for portfolio construction, central bank reserve policy, and future safe-haven research design, and should increasingly involve modelling gold alongside its competing macro-financial assets, and not in isolation..

Keywords: Gold, Safe Haven, Geopolitical Risk, Iran–Israel Conflict, MGARCH, Flight-to-Quality, Behavioural Finance, Efficient Market Hypothesis

1. Introduction

Gold is in a special place among commodities because it is used as an industrial commodity, as a personal commodity and as a monetary commodity. Central banks have been buying gold, gold demand from the retail sector has been strong, and gold-related financial products have increased the presence of gold in institutional portfolios (Yokus, 2024). Indeed, gold has always been considered the “canonical” asset of refuge in the event of geopolitical rupture, and has been for centuries, from the 1973 Yom Kippur War, to the 1990 Gulf War, and the 2022 Russian invasion of Ukraine. But this reputation has always been more historical than based on a firmly established, theoretically sound mechanism.

However, Madhavan (2023) showed in his study on India that gold did not serve the role of a safe haven in all the above mentioned crises, but rather shared the same movements as the equity market, and even against the equity market during COVID-19. The test case of Iran vs. Israel is a new one and has not been studied enough. In contrast to the crises in the existing literature, the conflict played out in a macrofinancial context of high policy interest rates, Fed talk of higher rates, a strong US dollar, and ongoing inflation worries related to energy-market disruption. This setup is very much what safe-haven theory does not have to explain: Past research has analyzed gold under two conditions: first, when the dollar and interest rates were relatively inactive; and second, when they were not analyzed at all. The present study is driven by this interaction, between a shock of geopolitical events and a tight monetary and currency regime, rather than by the geopolitical shock itself.

The empirical question is simple to state: Gold saw no price decline through the conflict, but it also did not increase in a steady monotonic way, as the classical “safe haven” story would suggest. Rather, the price chart reveals a rally and subsequent two corrective periods (late January and March 2026), despite the presence of high geopolitical risk. This condition is hard to explain with a pure geopolitical-risk view of gold prices, and instead suggests that other macroeconomic factors, the opportunity cost of holding a non-yielding asset as well as the liquidity premium on the US dollar and the varying expectations for rate cuts, are also part of the equation that determines gold’s path. This paper aims to formally test, using daily data over the conflict window, whether the traditional safe haven properties of gold continued, weakened or were structurally replaced by these macro-financial determinants and to interpret the results by providing an explicit set of financial theories in addition to mere description.

2. Literature Review

The literature addressing gold’s safe-haven property can be broken down into three major, somewhat conflicting, groups: (i) research that views gold’s safe-haven property as a structural characteristic; (ii) research that finds that the safe-haven property is dependent on the crisis, is only present in some crises, or is in decline; and (iii) research that views gold primarily as a function of macroeconomic fundamentals rather than crisis sentiment. The strands, however, read side by side and not chronologically reveal consensus and disagreement upon which this study is based.

2.1 Gold as a Stable Safe Haven

One of the earliest empirical regularities in this space was found by Capie, Mills and Wood (2005) who discovered that as the US dollar depreciates, gold prices tend to increase, which they argue is a consequence of gold being a dollar hedge for international investors. Demidova and Heidorn (2007) further expanded this conclusion and demonstrated that from 2000 to 2006 gold’s low correlation with other asset classes was not only a portfolio diversifier, but the best possible one. The stable haven perspective was backed up by Pandey (2023) who found that gold prices in India had increased 38 per cent during the COVID-19 period, reflecting increased demand for wealth preservation. The most cited theoretical work for the field was by Baur and Lucey (2010) who formally identified a hedge (an asset that is uncorrelated with stocks and bonds on average) and a safe haven (an asset that is uncorrelated or negatively correlated with stocks and bonds specifically during market turmoil). They applied this test to data from the United States, United Kingdom and Germany, and discovered that gold served as a hedge against equities and a safe haven during extreme stock-market stress, but that the safe-haven property of gold was short-lived and typically “worn out” after about fifteen trading days or so following the shock (Baur & Lucey, 2010).

2.2 Gold as a Conditional or Declining Safe Haven

The second string contradicts the stability suggested by the first string. In the case of India, Madhavan (2023) observed that the protective role played by Gold has been quite mixed in different crises, and in the case of COVID-19, the role of Gold was negative, as it co-moved with the negatively moving Equities. Yokus (2024) created a Gold Market Pressure Index from 1970 to 2023, showing that gold’s safe-haven qualities have come in three waves: the oil-shock and war period of the 1970s, and the period from 2009 to 2021, when the Global Financial Crisis, Arab Spring and COVID-19 occurred. Mondal, Ghosh and Sana and Paul (2025) studied Indian gold data during the Russia–Ukraine war and found that gold prices were more affected by negative news than by positive news of the same magnitude, which aligns with the tenets of investor loss aversion, and suggests that gold’s reaction to conflict is sentiment-dependent rather than simply geopolitical. Together, this strand supports the first with the added view that gold may exhibit this safe haven response, but not consistently for all types of crises and not across samples or over time — a point that is an object of study in its own right, when applied to the Iran–Israel case.

2.3 Gold as a Function of Macro-Financial Fundamentals

A third value is a minimum of geopolitical sentiment and presents a model where gold serves as a function of measurable macro variables. According to the modelling by Oxford Economics (2011) for the World Gold Council, a 10 percent increase in the US dollar corresponds to an estimated drop in gold prices of approximately 8.4 percent, with the dollar and real interest rates and financial stress the most significant factors. Comparing GARCH(1,1) and interest-rate-augmented GARCH-M, Tardiana, Akbar, and Widodo found that the latter model significantly outperforms the former model, with higher interest rates reducing gold returns due to their increasing opportunity cost of holding the gold asset. Neelan and Rani (2025) investigated the relationship between gold and the USD-INR exchange rate, the domestic equity index and the central bank’s repo rate for India spanning the years from 2019 to 2023 and identified strong positive correlation between gold and USD-INR exchange rate as well as gold and domestic equity index with 95 percent model accuracy, whereas the relationship between gold and repo rate of the central bank was found to be strong and negative. This strand’s results are consistent with Capie et al. (2005) for the dollar channel, but re-frames the question of gold pricing as one of macro-financial phenomena, with geopolitical risk as, at most, one of several inputs, a reframing that motivates the multivariate design pursued here.

Taken together, the three strands is unanimous that gold is not a purely safe haven, but as to which is the dominant force – the sentiment driven flight to safety or the opportunity cost/currency mechanics, they disagree. However, no existing study focuses on a conflict in which both mechanisms were likely operating at the same time, and in conflict with each other, increasing geopolitical risk driving gold higher, while a hawkish and high-yield monetary backdrop was driving gold lower. This is the particular contribution of the present study to the literature, the context being the Iran–Israel conflict, where the two forces can be estimated together, instead of being assumed away.

3. Theoretical Framework

The empirical design is based on five complementary theories, all of which address a different mechanism by which the conflict could have plausibly influenced gold prices, and goes beyond mere description.

3.1 Safe-Haven Theory

Baur and Lucey (2010) define a hedge as an asset uncorrelated in average sense with another asset class, and a safe haven as an asset uncorrelated or negatively correlated during stress periods specifically; a further distinction between a ‘strong’ safe haven (a significant negative correlation) and a ‘weak’ safe haven (a non-significant negative correlation) is further refined. This framework provides support for the null hypothesis tested in this study: Absence of a strong safe-haven effect for gold in the Iran–Israel conflict, the price of gold should be independent of, or inversely related to, deteriorating risk sentiment, whether in a normal or a monetary contractionary environment.

3.2 Portfolio Theory

The reason investors hold gold is that there is a correlation between gold and the other assets in a portfolio, and that correlation is usually low or even negative, making gold’s expected return unremarkable but meaning that its role in a portfolio reduces the overall portfolio variance. This logic is put into practice empirically with the finding by Demidova and Heidorn (2007) that gold has low co-movement with other assets, thereby enhancing the efficiency of asset portfolios. In contrast, even if the geopolitical story were bearish, as per Portfolio Theory, gold demand and thus price should increase whenever the correlation structure of the portfolio (apart from gold) deteriorates.

3.3 Flight-to-Quality Theory

In Caballero and Krishnamurthy (2008), flight to quality is a reaction to Knightian uncertainty, in which agents are unable to assign probabilities to outcomes, and therefore liquidate illiquid and risky assets and buy perfectly liquid and certain claims, which in today’s markets are overwhelmingly cash and US Treasury paper rather than physical goods. The theory directly contradicts the classic safe-haven theory that says that when a liquidity crisis is underway, capital tends to flock to dollar and Treasuries regardless of gold, since gold is not the same as a Treasury bill, which can be converted at a moment’s notice and for a nominal cost. This mechanism is central to interpreting why the US Dollar Index and Treasury yields might dominate gold pricing during the acute phase of a crisis.

3.4 Efficient Market Hypothesis (EMH)

The Efficient Market Hypothesis (EMH) of Fama (1970) suggests that market prices reflect all publicly available information, such as geopolitical events and signals from the monetary policy. Under EMH, gold’s price should quickly and fully respond to any news of a new stage in the escalation of the conflict, new guidance on interest rates, or new inflation data, and the next price change should be an adjustment to new information, not a continuation of a mispricing trend. These two short-term corrective swings are interpreted as being sensible reactions to new information regarding monetary-policy and profit, and not as a repudiation of gold’s intrinsic safe-haven quality.

3.5 Behavioural Finance

An alternative (narrowly rational) view is the prospect theory (Kahneman and Tversky 1979) and extension thereof to the commodity market, loss aversion: Investors find it more painful to lose a dollar than to gain one; the impact of bad news outweighs that of good news of the same economic magnitude. The authors of a recent paper by Mondal, Ghosh, Sana and Paul (2025) have applied this rationale to the case of gold during the Russia-Ukraine war, which they found to be strongly asymmetric. This is done using Behavioural Finance to explain the clustering of volatility and the sudden, transitory declines in the gold series, as indicative of herding and profit taking behaviours, but not of major changes in underlying fundamentals.

The five frameworks are not distinct: the empirical strategy below speaks to the Flight-to-Quality and opportunity-cost frameworks primarily, the MGARCH framework speaks to the volatility clustering as implied by Behavioural Finance, and the overall pattern of rapid price adjustment to the policy and profit-taking news is viewed through an EMH lens.

4. Research Gaps

• To date, no study investigates gold price dynamics from a daily data perspective during the Iran–Israel war (December 2025 – May 2026), and those few that do focus on older time periods (COVID-19, Russia–Ukraine war, and Eurozone debt crisis) lack the macro-financial context of high policy rates and hawkish Fed guidance that prevailed during the Iran–Israel war.
• There is very limited previous empirical literature on multivariate models of volatility (such as GARCH and TGARCH) or simple descriptive correlation analysis of gold alone. The purpose of this study was to combine the price-level determinants (via multivariate OLS) and the cross-asset volatility transmission (via MGARCH) in a single theoretically integrated approach, which is the combined approach that was followed in few studies.
• Existing safe-haven literature rarely explicitly compares empirical tests to competing theoretical models, such as the safe-haven theory, flight-to-quality theory, or behavioural finance; in this study, results are presented without pre or post-questioning the mechanism that the coefficients are actually associated with – that is, is it the safe-haven theory that the coefficients are consistent with, or is it the flight-to-quality theory, or is it the behavioural finance theory, etc.

5. Objectives of the Present Study

• Objective 1 — Macroeconomic Determinants of Gold Price Dynamics: To study the impact of macro-financial factors for the Iran–Israel conflict window (Brent crude oil, US Dollar Index, Federal Funds Rate, inflation, and 10-year Treasury yields) on gold price dynamics from 1 December 2025 to 15 May 2026.
• Objective 2 — Safe-Haven Asset Status: Testing the negative correlation of gold with opportunity-cost variables, as predicted by classical safe-haven pricing behaviour, during the conflict period to assess the relationship between gold and these macroeconomic indicators.
• Objective 3 — Explanation of Temporary Price Declines: To identify the macroeconomic and volatility-transmission mechanisms that explain the two temporary drops in gold prices which occurred during the conflict period and to assess whether these temporary drops contradict or just qualify the safe-haven asset theory.

6. Methodology

6.1 Research Design

A quantitative empirical time-series design is used to isolate gold price movement that is specific to the local and conflict area from the macroeconomic drivers during the period under study. The theoretical basis for the design choice outlined above is that both Flight-to-Quality Theory and opportunity-cost logic for holding a non-yielding asset make testable predictions about the sign and magnitude of the relationship between the price of gold and the price of the dollar, interest rates, and Treasury yields; therefore, it is necessary to use a multivariate regression framework rather than a univariate event-study analysis of gold alone to adjudicate among competing explanations for the same price path.

6.2 Study Period, Variables and Data Sources

To capture the full story of the escalation, from the initial shock and the peak of the conflict to the start of de-escalation, the conflict window of 1 December 2025 to 15 May 2026 was chosen, and was sufficiently short that the estimated relationships are likely to be the conflict regime and not some longer-run structural relationship that would take multiple monetary cycles to identify. A shorter window will lead to inadequate observations for GARCH estimation, while a much longer window will wash out the conflict related signal by adding unrelated macro regime changes, which is mentioned as a limitation in Section X.

The dependent variable for this analysis is the Gold’s Closing Spot Price (XAU/USD, USD per troy ounce). The following are the variables chosen for their theoretical relevance to different channels: Brent crude oil, the trade-weighted US Dollar Index, the Federal Funds Rate, US CPI inflation, and 10-year Treasury yields, as each channel relates to a different aspect of the gold’s theoretical appeal. As done in the macro-finance literature reviewed, above (e.g., Tardiana et al., Neelan & Rani, 2025), data were sourced from Investing.com (gold, oil, Dollar Index) and the FRED database (Federal Funds Rate, CPI, 10-year Treasury yield).

6.3 Data Transformation and Justification

All assets were changed to continuous time log returns: ln (Pₜ / Pₜ₋₁); the Federal Funds Rate was first-differenced: ΔYₜ = Yₜ − Yₜ₋₁; and inflation was log-differenced. The transformations are not willy-nilly: most macro-economic and asset-price series are not stationary in levels and regressing one that is non-stationary on another runs the risk of spuriousness, of having an R² that is too high and t-statistics that are too high owing to common trends. Therefore, the tests following the ADF test, such as regression and covariance models, require stationarity of the data before they can be validly applied, and the ADF test is performed before these tests.

6.4 Data Analysis Techniques

Four successive techniques were used and each was chosen because it is uniquely suited to test something:

1. Descriptive Statistics set the initial baseline distributional profile of gold prices (mean, dispersion, skewness, and kurtosis); they also provide an initial, model free check in regard to whether the price series exhibits the fat-tailed, panic-driven behaviour that would typically result from a strong safe-haven reaction.
2. The Augmented Dickey-Fuller (ADF) Unit Root Test is used to test stationarity of each transformed series, under the null hypothesis of a stochastic trend and the alternative hypothesis of a mean-reverting, stationary process. This step is a methodological requirement because of the Gauss-Markov assumptions that underlie OLS and because of the results obtained in Section VII, which found that all of the series tested were stationary before being regressed, thereby removing the spurious-regression risk mentioned.

                                        Δyt​=β1​+β2​t+δYt−1​+i=1∑k​αi​ΔYt−i​+εt​

3. The specific and falsifiable signature that the Flight-to-Quality Theory and the opportunity-cost mechanism predicts is a significant negative coefficient on the interest rate or the Treasury yield, and the specific signature of currency-driven liquidity preference is a significant negative coefficient on the Dollar Index, which is the multivariate OLS Regression model that directly operationalises these theories.

               Goldt​=β0​+β1​(Oilt​)+β2​(Dollart​)+β3​(InterestRatet​)+β4​(Inflationt​)+β5​(Treasuryt​)+εt

4. The OLS price-level regression is unable to identify time-varying conditional covariance between gold and any other competing asset on its own. Multivariate GARCH (MGARCH) Covariance Modelling addresses this issue. This is because it shows that two different and conceptually unrelated questions are being answered: Does the price level move with a particular variable (OLS)? and Does the risk level co-move with the price level of that particular variable (MGARCH)?

6.5 Robustness Consideration: Multicollinearity Diagnostics

The Federal Funds Rate is macroeconomically related to both the Treasury yield and the Dollar Index, both of which are included in the regression; when the Fed is hawkish, mechanically, the yield will rise and the Dollar Index will appreciate, which means that the regression is potentially subject to multicollinearity. This study’s OLS diagnostic output indicates a condition number of 1.60 × 10⁵, a value that is generally considered to be a sign of considerable multicollinearity among the predictors. It does not change the estimated signs and significance of the coefficients (which are predicted by the theoretical framework independently); rather, it indicates that the magnitude to be attributed to any coefficient, particularly the Federal Funds Rate coefficient, should be understood as that of a channel of tightening and of liquidity preference, instead of as purely a variable-specific effect. This is reported as an interpretive caveat on the existing output of the OLS rather than as a reason to expand the data set and is revisited in Section X with a recommendation that studies conducted in the future should use Variance Inflation Factor screening or a VAR/VECM model that explicitly models the interdependence among these predictors.

7. Results

7.1 Baseline Descriptive Statistical Profile

Table 1 presents the distributional properties of gold prices across all 166 observations.

Table 1. Descriptive Statistics — Gold Price (USD / troy oz), December 2025 – May 2026

The mean gold price across the window was $4,714.69 per troy ounce, ranging between $4,189.05 and $5,400.25. The skewness is slightly positive (0.1737) which suggests a slightly longer right tail toward the end of the initial rally phase, and the negative kurtosis (−0.8225) suggests a flatter, thinner distribution than a normal distribution — a platykurtic distribution. From an economic point of view this is an interesting and somewhat surprising result since a true ‘panic’ safe-haven rally would typically produce fat right tails, i.e. a few particularly large daily gains. Their absence indicates that the price of gold over the window increased slowly and in a widely distributed manner, through many relatively small daily moves, rather than through a small number of large, panic-related price jumps — a precursor, backed by the regression result below, of the fact that gold’s price action was fueled as much by macro-financial pressure as by the occasional extreme geopolitical shocks.

7.2 ADF Unit Root Verification

Table 2. Augmented Dickey-Fuller (ADF) Unit Root Test Results

All six series have been found stationary, and the linear multivariate framework used is free of the risk of spurious regression by virtue of the null hypothesis of no unit root being rejected. This satisfies the precondition in Section VI.C – to move on to OLS and MGARCH stages.

7.3 Multivariate OLS Parameter Estimation

Table 3. Multivariate OLS Regression Results — Gold Price vs. Macroeconomic Predictors

Model diagnostics: R² = 0.711 | Adjusted R² = 0.702 | F-statistic = 78.57 (p = 2.77 × 10⁻⁴¹) | Jarque-Bera = 0.902 (p = 0.637) | Durbin-Watson = 0.360.

These 5 macro-financial variables account for 71.1% of the price movement of gold, which is a satisfactory level of fit for a model with 166 daily observations. From the point of view of economy, each coefficient corresponds to a particular mechanism in the theoretical model of Section III. The Dollar Index coefficient of −125.94 implies, for every 1 point increase in the DXY, gold’s price in dollars declined by approximately $125.94, all else equal, which is the hallmark of the Flight-to-Quality effect that favors transactional USD liquidity over bullion. Both the Federal Funds Rate coefficient (minus5,907.13) and the Treasury yield coefficient (minus1,630.40) measure the opportunity-cost mechanism: when the opportunity cost of holding gold increases, due to an increase in the competing opportunity cost of holding gold-free securities, investors reallocated away from gold and toward the other securities. The positive oil coefficient (9.20) represents the classical energy-shock-to-inflation-expectations channel but its contribution to the path of gold’s price path was relatively small compared to the dollar and yield channels, the oil’s estimated size was not as large, and the other channels were also negative in size. Importantly, inflation’s insignificance (p = 0.303) is another important economic discovery: gold did not behave as the classical ‘inflation-hedge’ asset over this six month period, more than likely reflecting the macro-fundamentals strand of the literature (Oxford Economics; Tardiana et al.). The small average Durbin-Watson statistic (0.360) suggests positive serial correlation in daily residuals, which is typical of financial data, where high volatility is frequently accompanied by high returns, as described in the MGARCH stage that follows.

7.4 Multivariate GARCH (MGARCH) Covariance Modelling

Table 4. MGARCH Conditional Covariance Matrix — Gold vs. Competing Asset Classes

The MGARCH results include a new dimension of risk, which cannot be accounted for by the OLS price-level model. The price level of gold and the Treasury yield exhibits the highest (in absolute terms) negative conditional covariance (−0.000866): gold’s price level turns out to move counter-cyclically to the yields, and so do the gold and Treasury risk (respectively, their volatility levels) – the latter is the volatility-level version of the former, for the “Flight to Quality” mechanism. Compare this with the negligible volatility interaction of gold and interest rates, (−0.000007) and gold and inflation, (+0.000004): in both cases, the largest OLS price-level coefficient is in the interest rate, which is the single largest among all OLS price-level coefficients. Taken together, this suggests that interest rates had a strong influence on gold primarily via the opportunity-cost channel in the price level, whereas they had a strong influence on gold also via the active, short-run risk-transmission channel — which is directly relevant to how quickly the effects of each channel were expected to be felt.

7.5 Interpreting the Two Price Declines

The elevated path of the gold price is interrupted by two identifiable corrective episodes, a steep one around 29 January 2026 and a longer one from 11 March 2026 to 23 March 2026. An Efficient-Market interpretation would be that the January episode was due to profit-taking after a quick price pickup and not a re-evaluation of the geopolitical risk itself (Reuters; The Guardian), as most of the reporting was contemporaneous. The March episode was consistent with a more hawkish given set of Federal Reserve signals, a strengthening Dollar Index and higher Treasury yields in the wake of continued inflation concern (Reuters; StoneX), and thus fits in perfectly with the negative Dollar Index, interest rate and Treasury yield coefficients estimated in the OLS model. Baur and Lucey (2010) distinguish the following: In both episodes, gold was not necessarily losing value in the sense of being negatively correlated with risk assets when a financial crisis occurred, it was only that the price level of gold was temporarily dominated by another macro-financial channel, in addition to, and not instead of, the geopolitical premium.

8. Discussion

The central finding – that gold’s safe-haven attributes during the Iran–Israel conflict were both real and conditional – expands and contradicts in part the previous literature reviewed in Section II. It establishes a similar but macro-led decay pattern that is ongoing throughout the six-month conflict, and in which the geopolitical premium of gold has been more than offset, and at times even dominated, by monetary tightening and dollar strength, rather than by time alone, as Baur and Lucey (2010) noted in their short-lived safe-haven finding.

It also addresses the apparent contradiction between the findings of Madhavan (2023) that gold did not prove to be a safe haven or diversification asset in various crises in India, and those of Pant (2023) and Demidova and Heidorn (2007) that it demonstrated strong safe haven and diversification benefits. The present results indicate that the disagreement may not be about the safe-haven role of gold itself, but that it may relate to the prevailing interest rate environment and the currency’s relative strength at the time of the different crises (COVID-19, where Madhavan found the safe-haven role of gold was weakest, and high interest rates, strong dollar in Iran–Israel in 2026). Differences in the monetary-policy regimes of studies, not conflicting evidence regarding gold itself, may account for much of the disagreement in the broader literature found here if opportunity cost and currency strength prove to be the predominant swing factors.

Despite the vastly different setting, crisis, geography and period, the results are very similar to the macro-fundamentals strand of the literature on the dollar and interest-rate channels (Oxford Economics; Tardiana et al.; Neelan & Rani, 2025), yielding the same negative-coefficient pattern. This study differs from the strand cited above in its very strong positive oil coefficient—this is a co-signal of oil forcing gold’s price level up even though the two assets’ volatility profiles diverged—and a negative Gold-Oil MGARCH covariance, which suggests that oil was a supply-shock safe-commodity that was not primarily an inflation pass-through, as the purely macro-fundamentals literature does not typically test for a nuance that this study’s design necessitates, given their separate focus on either price-level or covariance models.

Last, one needs to bear in mind that inflation is insignificant in comparison to the basic Feldstein (1978) thesis that gold protects against inflation because its quantity of real gold is fixed. During that six-month period that relationship didn’t hold statistically, it makes sense that the expectation of inflation would be priced into the interest rate and Treasury-yield movements, as would be expected by the Efficient Market Hypothesis, which states the bond-yield market is an efficient market and that inflation would be expected to manifest indirectly, through yields, rather than as a separately significant factor in gold once one already includes the bond yields.

9. Implications of the Study

The study provides both theoretical and practical insights into the interpretation of the part played by gold in geopolitical crisis situations. They argue that, in theory, the crisis behaviour of gold has only been described in the context of Safe-Haven Theory, and that a complete description of gold’s behaviour needs to be nested within Flight-to-Quality Theory and opportunity-cost reasoning; the same crisis can have both higher aggregate demand for gold and higher demand for competing ‘safe’ assets, which is principally the US dollar and Treasuries, and when policy rates and yields are already high, these alternative assets’ price path dominates that of gold.

The findings are concerning for investors and portfolio managers who might want to use gold allocation decisions as a reaction to geopolitical news. The Federal Funds Rate, Treasury yields, and the Dollar Index each accounted for most of the price action in this conflict, so allocation and hedging around gold during future geopolitical market events should take into consideration more than just the intensity of the event itself but also the overall monetary-policy stance and currency trend in existence.

The negative Gold-Treasury covariance implies that gold and sovereign debt served as substitute safe assets during this conflict, not complementary ones, and thus may have important implications for the way reserve managers view the diversification benefits that gold and sovereign debt provide during concurrent geopolitical and monetary stress.

10. Limitations of the Study

1. Limited Observation Window and Sample Size

The 166 observations over six months reflect the short-term market reaction but are too few to conclude the stability over the longer-term of the estimated relationships, which are known to vary from one economic regime to the next and for which evidence of long-term stability would be more persuasive with a longer window that covers the period before, during, and after the conflict.

2. Restricted Econometric Scope

The research methods used are Descriptive Statistics, ADF testing, multivariate OLS and MGARCH. Such methodologies as more dynamic frameworks — Vector Autoregression (VAR) and Vector Error Correction Models (VECM), Impulse Response Functions and formal event-study methodology — were not applied and could provide more precise information with respect to the direction of causality and the speed of shock transmissions; the multicollinearity flagged, however, through the OLS condition number (Section VI.A particular case in point is the case of E) which would be suitable for using a VECM or an explicit VIF-screened specification as a suitable complement to the current results, rather than a replacement.

3. Omitted-Variable Risk

The list of the five macro variables chosen is not comprehensive of possible drivers of gold prices. The investor sentiment indices, central bank gold-purchase activity, liquidity conditions in the markets, the speculative positioning and the expectations regarding exchange rates were not included, and the exclusion of these variables may lead to a bias of the estimated regression coefficients when they are correlated with the included regression variables.

11. Conclusion

This study analyzed whether gold maintained its traditional safe-haven role throughout the Iran–Israel conflict (2025–2026) based on the 166 daily observations and using a 4-stage econometric design based on five complementary financial theories. The evidence is used to provide a conditional answer rather than an absolute. Gold showed a consistent movement as the war began, reacting as a traditional safe haven as worries grew. As the war continued, however, the US Dollar Index, the Federal Funds Rate and Treasury yields had statistically significant negative influence on the price level of gold, and gold and Treasuries were found to be volatility substitutes more than co-movers in the MGARCH estimates, which is a clear indication of Flight-to-Quality Theory working concurrently with, and at times against, the safe-haven mechanism. The inflation effects had no statistically detectable impact on the period, with oil having a small positive effect.

Read the theoretical framework in Section III, gold is not a safe haven that failed, it is a safe haven that is only secure when the dominant monetary/currency regime is one that is based on gold. This reinterpretation makes it easier to explain part of the apparent divide in the literature between those studies that have found strong safe-haven behavior and those that have found little or no such behavior: the distinction might not be so much a distinction between gold as a safe investment versus not, but rather between interest rates and the value of the dollar during the crisis. Further tests and refinements of this conditional account of the role of gold as a modern safe haven asset could be carried out in future research that further extends the observation time window, takes into account the role of VAR/VECM causal frameworks, and explicitly models the interest-rate regime as a moderating variable.

12. References

1. Baur, D. G., & Lucey, B. M. (2010). Is gold a hedge or a safe haven? An analysis of stocks, bonds and gold. Financial Review, 45(2), 217–229.
2. Baur, D. G., & McDermott, T. K. (2010). Is gold a safe haven? International evidence. Journal of Banking & Finance, 34(8), 1886–1898.
3. Bialkowski, J., Bohl, M. T., Stephan, P. M., & Wisniewski, T. P. (2014). The gold price in times of crisis. International Review of Financial Analysis.
4. Caballero, R. J., & Krishnamurthy, A. (2008). Collective risk management in a flight to quality episode. The Journal of Finance, 63(5), 2195–2230.
5. Capie, F., Mills, T. C., & Wood, G. (2005). Gold as a hedge against the dollar. Journal of International Financial Markets, Institutions and Money, 15(4), 343–352.
6. Demidova-Menzel, N., & Heidorn, T. (2007). Gold in the investment portfolio. Frankfurt School of Finance & Management Working Paper.
7. Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. Journal of Finance, 25(2), 383–417.
8. https://www.nber.org/papers/w0296
9. Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291.
10. https://www.iima.ac.in/sites/default/files/2023-06/Sruthy%20Madhavan.pdf
11. Markowitz, H. (1952). Portfolio selection. Journal of Finance, 7(1), 77–91.
12. Mondal, R., Ghosh, R., Sana, A., & Paul, B. (2025). Measuring the volatility in gold prices of India during the Russia–Ukraine crisis: Evidence from the TGARCH model. International Journal of Management and Human Science, 9(2), 59–65.
13. Neelan & Rani et al. (2025) https://www.ijcrt.org/papers/IJCRT2502590.pdf
14. Oxford Economics. (2011). The impact of inflation and deflation on the case for gold. World Gold Council commissioned study.
15. Pant, M. (2023). The performance of gold during COVID crisis. International Journal for Research Trends and Innovation, 8(12).
16. Tardiana, A. L., Akbar, H., Firmansyah, G., & Widodo, A. M. (2024). Integration of GARCH models and external factors in gold price volatility prediction: Analysis and comparison of the GARCH-M approach. Eduvest — Journal of Universal Studies, 4(5), 4011–4023.
17. Yokus, T. (2024). Definition of world gold price crises: Gold price crises from January 1970 to December 2023. Proceedings of the 8th International Izmir Economics & Business Administration Congress, 13–22.

Additional Sources and Dataset References

1. Reuters — Gold falls as investors take profits after record highs.
2. Reuters — Gold slips on stronger Dollar and delayed rate-cut expectations.
3. The Guardian — Precious metals reverse after record surge.
4. StoneX — Gold prices fall as liquidity pressures override safe-haven demand.
5. World Gold Council Research and ETF flow data.
6. IMF Global Financial Stability Reports.
7. Investing.com — Gold Spot (XAU/USD), Brent Crude, and US Dollar Index historical data.
8. Federal Reserve Economic Data (FRED), Federal Reserve Bank of St. Louis — Federal Funds Rate (FEDFUNDS), Consumer Price Index (CPIAUCSL), and 10-Year Treasury Yield (DGS10).
9. com. (2026). Gold Spot (XAU/USD) Historical Data — Daily Closing Prices. Retrieved from: Gold Spot (XAU/USD) Historical Data – Investing.com
10. Federal Reserve Economic Data (FRED). (2026). Crude Oil Prices: Brent – Europe (DCOILBRENTEU). Federal Reserve Bank of Louis. Retrieved from: FRED Brent Crude Oil Prices (Europe)
11. 11. com. (2026). US Dollar Index (DXY) Historical Data — Daily Closing Values. Retrieved from: US Dollar Index (DXY) Historical Data – Investing.com
12. Federal Reserve Economic Data (FRED). (2026). Federal Funds Effective Rate (FEDFUNDS).Federal Reserve Bank of Louis. Retrieved from: FRED Federal Funds Rate (FEDFUNDS)
13. Federal Reserve Economic Data (FRED). (2026). Consumer Price Index for All Urban Consumers: All Items (CPIAUCSL). Federal Reserve Bank of Louis. Retrieved from: FRED CPIAUCSL Inflation Dataset
14. Federal Reserve Economic Data (FRED). (2026). 10-Year Treasury Constant Maturity Rate (DGS10). Federal Reserve Bank of Louis. Retrieved from: FRED 10-Year Treasury Yield (DGS10)