India's demand for gold: some issues for economic development and macroeconomic policy.

Kannan, R. ; Dhal, Sarat

Abstract

India is the largest consumer of gold in the world. Gold is a major vehicle of saving for large number of low and middle income households in rural and urban areas. This study provides analytical and empirical perspectives while identifying the key drivers of India's gold demand during the period 1980-2005. Broadly, empirical findings suggest that India's gold demand is significantly influenced by real income and a set of variables pertaining to monetary, fiscal and financial sector policies such as interest rate, exchange rate, personal income tax, government spending to ease economic and social uncertainty and wealth (asset price), besides the relative price of gold. These findings have critical implications for development policy, financial intermediation and gold market in the Indian context.

JEL Classification: D1, D4, D120, D110, E21, C32,

Keywords: Gold, Consumption, Savings, Demand, Time series models (Error Correction Model)

I. INTRODUCTION

Though gold no longer commands the role of a monetary standard, it continues to remain attractive for its consumers, producers and investors across the globe. In India, gold has immense value to the society in terms of jewellery consumption for adornment as well as a major vehicle of wealth accumulation (asset demand) by large number of low and middle income households in rural and urban areas. India is the largest consumer of gold, accounting for a fifth of annual global gold consumption. The gold industry in India employs about half a million people with a turnover more than US $20 billion. Almost all of India's gold demand is met by imports. Thus, for India's gold market, gold demand is the mainstay.

There exists a considered view, attributable to Lord John Maynard Keynes, that India's gold consumption reflects the 'ruinous love of a barbaric relic'. One can only imagine as to how Lord Keynes would have reacted to the surge in India's gold demand in an era of rapid economic progress during the reform period. Gold demand remains unabated, despite the sharp increase in gold prices. Several pertinent questions arise. Why does India consume so much gold? Should the surge in gold demand be construed as socio-economic backwardness? Is gold demand affected by several other factors, besides the gold price? Answers to these questions warrant analytical and empirical perspectives with a focus on principal determinants of gold demand. From policy perspective, gold demand has linkages with various macroeconomic aggregates (Reddy, 1997). Gold demand assumes critical importance for 'inclusive growth', which is currently the major objective of national planning in India and 'financial inclusion' entailing adequate access of people to the organized financial system for achieving 'inclusive growth' (Mohan, 2005). Since independence, concerted efforts have been made toward financial development under public sector banks. While rural areas account for about 70 per cent of total bank branches, their share in aggregate deposits is only 30 per cent. Although the buoyant gold demand exhibits saving potential of people, but at the same time a strong alternative to financial instruments including bank deposits and risk free government bonds. We view that it is appropriate to go beyond economic backwardness perspective and recognise India's gold demand as the saving motive of people in response to macroeconomic policy and the evolving social and economic environment.

Against the above backdrop, the study is thus organized. Section 2 provides a brief review of India's gold policy. Section 3 dwells on an analytical perspective on the motives for holding gold and key drivers of gold demand deriving from the literature, as well as the role of macroeconomic policies in the Indian context. Section 4 outlines the methodological framework and data used by the study. Section 5 presents stylized facts about India's gold market, followed by empirical findings in Section 6. Policy perspectives are set out in Section 7 and Section 8 concludes.

II. INDIA'S GOLD POLICY

A study on India's gold demand would not be complete without a discussion on gold policy, which forms an integral part of macroeconomic policy. Broadly, there is a shift in the approach of policy from a plethora of restrictions to a liberalised gold market during the reform period as compared with earlier periods. The policy approach no longer contends with whether gold is productive or unproductive asset. The role of a liberalized and developed gold market in the interest of consumers is being increasingly realised and efforts are underway for integrating the gold market with financial markets (Reddy, 1997, 2002, Thorat, 1997, Bhattacharya, 2002).

In retrospect, during the British rule, gold inflows and outflows were unrestricted mainly due to the practice of metallic standards. Restrictions on gold imports were imposed only during the two world wars. After independence, Government of India adopted a fully restricted policy toward gold. Saving and investment in gold were viewed as a loss of precious foreign exchange resources, with adverse consequences for economic development and technological progress (Bhattacharya, 2002). Accordingly, the evolution of the gold policy until economic reforms in the early 1990s centred around the major objectives of weaning away people from gold, reducing domestic demand, regulating supply of gold, and domestic prices, curbing smuggling and black income and conserving foreign exchange (Reddy, 2002). Accordingly, three major strategies encompassing legislative measures, control over domestic production and gold mobilization schemes were adopted. The Foreign Exchange Regulation Act (FERA) of 1947 and its successive legislation in 1973 and Gold Control Act 1968 empowered government to restrict exports and imports of gold. In 1956, the country's only gold mines in the southern region was nationalised to attain government control over gold production. In the same year, the proportional reserve system (under which a minimum proportion of the currency liability was required to be covered by foreign assets including gold) was abolished and replaced by a minimum absolute level for foreign assets for the Reserve Bank of India (RBI) for currency issue purposes, thus, debasing the monetary role of gold in India. Government of India attempted various gold bond schemes over the years in 1962, 1965, 1980, 1993, and 1999. However, these schemes achieved little success (Bhattacharya, 2002). An interesting feature of the gold control regime was that it prohibited all transactions between residents and non-residents. This coupled with administrative complexities resulted in the failure of the Gold Control order, as smuggling of gold and unofficial transactions in foreign exchange flourished.

In the wake of balance of payment crisis in 1990-91, external sector policies in general and gold policy in particular witnessed a major review. The High-level Committee on Balance of Payments 1991 (Chairman: Dr. C. Rangarajan), Government of India, examined issues relating to valuation of Gold held by the central bank, external commercial borrowing against gold and gold bond scheme. The Committee felt that caution should be exercised in creating permanent institutions or corporate bodies for mobilising gold. Gold bonds may be launched immediately as an experiment and a review be made of the economics of such a scheme. The first major policy reversal occurred in 1990 with the repeal of the Gold (Control) Act. Subsequently, in 1993, the provisions of the FERA, 1973 relating to gold were repealed. In order to augment supply through official channels, non-resident Indians (NRI) were allowed to bring in 5 kg of gold as part of their baggage (subsequently raised to 10 kg) every six months subject to payment of import duty, which was increased in the early 1990s. The other channel for importing gold was by way of issuance of special import licenses (SIL) for jewellery export purposes only.

By the mid-1990s, there was growing realisation that the negative attitude towards gold on the presumption of productive or non-productive savings was inconsistent with liberalised framework for economic progress. A positive approach, on the other hand, entails significant reduction in transaction cost or the reduction in difference in price between world markets and Indian gold market after liberalization (Reddy, 1997). In line with this changing perspective, the Committee on Capital Account Convertibility 1997 (Chairman: Shri S.S. Tarapore) recommended that free import of gold should be a pre-condition for further reforms in the external sector. For developing domestic gold market, the Committee made recommendations for allowing banks to freely engage in gold business in domestic and international markets. Thus, gold banking followed by extending authorization to commercial banks to import gold for sale or loan to jewellers and exporters. This policy has succeeded to a large extent in curbing illegal operations in gold and foreign exchange markets and reducing the disparity between international and domestic prices of gold from as high as 50 per cent during the 1980s to about 2 per cent in recent years. In continuation of liberalised gold policy, Government of India lifted the forty-year-old restriction on gold derivatives in 2004, allowing gold futures trade in the commodity exchanges. Following this initiative, in a span of two years, gold surpassed all other commodities for futures trading in the Multi-Commodity Exchange and National Commodities and Derivative Exchanges.

III. DETERMINANTS OF GOLD DEMAND

Consumers and investors hold gold for various motives, deriving from the unique physical attributes of gold such as scarcity, safety, and liquidity. Gold is endowed with the scarcity value due to limits to under-ground level stock and mining capacity. Gold is fungible and indestructible (Lawrence 2003). It has universal acceptance and provides a liquid form of saving as compared with various other physical and financial assets. Gold economists also identify a host of macroeconomic and financial variables, which impinge on gold demand (Abeken, 1983, Aggarwal and Soenen, 1988, Capie et al. 2004, Ghosh et al. 2002, Lawrence, 2003, Klapwijp, 2003, Mani and Vuyyuri, 2003, Patel, 1950, Ranson and Wainwright, 2005 Sharma et al. 1992 and Vaidyanathan 1999). However, economists offer alternative perspectives as to how these factors would impinge on gold demand as discussed in the following.

First, there exists a bi-polar view on income and gold consumption relationship. Classical economists viewed gold jewellery as luxurious consumption, driven by higher level of real income and economic prosperity of people. On the contrary, the Keynesian school viewed gold consumption as socio-economic backwardness, i.e., the inverse relationship of gold consumption with economic progress. Empirical studies also provide a mixed perspective. Ghosh et al., (2002) and Lawrence (2003) refute any significant effect of real income on gold demand in advanced economies such as the US, contrary to the findings of Sherman (1983) and Baker, et al. (1985) for industrial countries. Patel (1950), Sharma et al., (1992), Rao and Nagabhushanam (1960) and Vaidyanathan (1999) recognise real income as a significant driver of gold demand in India.

Second, according to standard demand theory, quantity demand for goods and services should be negatively influenced by prices. Economic theory also recognises an exception to this rule, a la, 'Veblen's effect' relating to luxurious goods. For the latter, higher prices enhance the value of consumption and thus, induce greater demand. In reality, gold prices may affect gold demand variously, depending upon jewellery fabrication for first use and consumption or saving motive (Pulvermacher, 2005). For low and middle-income countries, higher real gold price may imply greater costs for holding gold asset for retail consumers and savers. It is advocated that the relative price or the intrinsic value of gold (the price of gold in relation to prices of other consumer goods) should affect demand, since consumers have to optimize between consumption of gold and a host of other goods (Patel, 1950, WGC, 2005). Empirical studies in general find evidence of negative price elasticity for gold demand (Patel 1950, Rao and Nagabhushanam, 1960, Feldstein, 1980, Vaidyanathan 1999). Mani and Vuyyuri (2003), however, showed that lagged gold price, attributable to the expectation of higher price, had significant positive influence on gold demand (current gold price) in India.

Third, exchange rate affects gold price in domestic currency, since commodity prices are mostly quoted in terms of the US dollar, the major invoicing currency for world trade (Sjaastad and Scacciavillani, 1996 Kavalis, 2006). In this context, it is postulated that gold serves as the hedge against changes in a currency's value both internally against inflation and externally against other currencies (Chua and Woodward, 1982, Capie, et al., 2004). Exchange rate may also be construed as an asset price for the wealth holder, especially the institutional investors, in an open economy and affect investment demand for gold (Smets, 1997).

Fourth, as an alternate instrument of saving, the return on gold may be compared with the return on risk-free instruments such as bank deposits and government bonds. Greater the opportunity cost, lower the propensity of saving in gold (Reddy, 2002). In a climate of low interest rates, gold provides an attractive investment for people who believe that inflation will pick up (Dalglish, 1993). Accordingly, real interest rate is expected to influence gold demand. In the Indian context, rural and urban households have different perceptions towards savings in bank deposits, treasury instruments and gold. In rural India, people are yet to be fully reliant on commercial banking sector, reflecting the nature of access to banking services and socio-economic-cultural factors influencing their preference to financial and physical savings. Gold serves as a 'safe haven' for precautionary saving motive, providing a strong alternative to interest bearing deposits. Saving in bank deposits and government bonds is largely confined to the urban middle class including salary earners, as they are sensitive to interest income. Moreover, until a year ago, salary earners were allowed income tax rebate on contractual savings (national savings certificate, provident funds etc). Interest on contractual savings has been reduced by about 300 basis points while tax rebate has been withdrawn.

Fifth, equity securities provide another alternative for savers and investors. Generally, equity prices are used for capturing the wealth effect on consumption and saving/investment (Shirvani and Wilbrate, 2000). Capital gains arising from an increase in equity prices in the medium-longer horizon may entice consumers for greater consumption of gold. This implies that gains from financial assets have to be translated into less risky real assets such as gold (3). Empirical evidence on the effect of equity prices on gold market remains inconclusive (Pulvermacher, 2004). In the US, Lawrence (2003) shows that returns on gold are less correlated with returns on equity and bond indices, attributable to portfolio diversifier role of gold. Similarly, Smith (2001) finds small and negative short-run correlation between returns on gold and the US stock price. In India, Mani and Vuyyuri (2003) reveal insignificant effect of the stock price on the domestic gold price.

Sixth, as a precious metal with scarcity value, favourable supply condition may induce greater accumulation of gold. The supply condition may be captured through changes in gold mining production. However, in an environment of increasing sales by central banks, it may be appropriate to take into account the variation in aggregate gold supply.

Seventh, the relationship between savings and uncertainty is also extended to gold (Abken 1980, Koutsoyiannis, 1983). Economic and social uncertainty may be attributable to a situation when people find difficulties in being engaged in productive work on a sustained basis. Economic and social uncertainty may also be related to government spending on social security provisions and productive infrastructure. In western countries, low propensity of aggregate saving and gold consumption may be attributable to large government spending on social security. In developing economies, though governments spend on provisions of social security, it could be inadequate, leading to private arrangements for precautionary savings. In the Indian context, appropriate data on employment or unemployment are not available. Thus, the empirical analysis of the study relies on proxy variables such as government's development or capital expenditure to exemplify the impact of economic and social security on gold demand.

Additionally, various other indicators have been used by empirical studies on gold. These include demand for substitutes like silver and other metals (Koutsoyiannis, 1983, Sharma et al. 1992, Ghosh et al., 2002), aggregate financial savings (Vaidyanathan, 1999) and agricultural income (Sharma et al., 1992). It is to be noted that some of these variables could be related to the principal determinants of gold demand discussed in the above. Illustratively, financial savings would be highly correlated with income, wealth and interest rate. As regards substitutes, gold price may have high correlation with price of its substitute such as silver and thus, pose problems for the empirical estimation of gold demand function.

Role of Macroeconomic Policy: the Indian Context

In the Indian context, monetary, fiscal and financial sector policies have played a major role in influencing the paths of economic growth, inflation, interest rates, exchange rate and asset prices, which serve as key drivers of gold demand. Until 1991, India's macroeconomic policy environment was characterised, broadly, with fiscal dominance and financial repression, attributable to the objective of minimising interest cost to the treasury, rigidity in the bank rate, the policy rate at which central bank lends to government, statutory pre-emption of banks' funds using high level of reserve and liquidity requirements and regulation of interest rates on banks' deposits and credit. The external sector remained closed with restrictions on capital inflows and exchange rate pegged to a basket of currency to promote exports and curtail imports. Financial system lacked innovations with dominance of public sector banks. Deposits and government securities remained principal instruments of financial savings. Financial markets relating to equities and insurance existed but they lacked growth momentum.

As part of financial sector reform and liberalised external sector in the 1990s, the operating framework of monetary policy witnessed a shift in the focus from direct control over credit and interest rates to moderation of reserve requirements and freedom of banks to set interest rates [3]. Current account of balance of payment was made convertible and exchange rate was allowed to be market determined. Capital account was gradually liberalised to encourage private capital flows from abroad. Initially, faced with the balance of payment crisis and hardening of inflation condition, Reserve Bank of India raised the policy rate by 200 basis points to 12 per cent in 1991 from 10 per cent in the 1980s, besides increasing the reserve and liquidity requirements to contain money growth. Exchange rate was devalued in two stages before allowing it to be market determined. Later, the policy rate declined by 600 basis points during the second half of the 1990s, since inflation receded to the targeted 5 per cent (10.5 per cent during the early 1990s and 7 per cent during the 1980s). In tandem with the policy rate, interest rates on bank deposits also increased by 200 basis points, rising from an average of 10 per cent in the 1980s to 12 per cent during first half of the 1990s, in order to step up financial savings for accelerating economic growth. Similarly, the yield on government bonds went up by 300 basis points from an average of 9 per cent in the 1980s to 12 per cent in the first half of the 1990s. Thereafter, due to moderating inflation and increasing focus on strengthening banks' balance sheet, deposit interest rates came down sharply to about 5 per cent. Similarly, the yield on government bonds declined to 6.6 per cent during 2002-05 from as high as 13 per cent during the first half of the 1990s. In real terms, deposit interest rates, which averaged at 4 per cent during the first half of the 1990s, declined to 0.8 per cent during 2002-05. Real yield on government bonds turned marginally negative at 0.2 per cent during 2002-05 as compared with positive 4 per cent during the first half of the 1990s. A critical feature of banking sectors' interest rate policy during the 1990s is that with a view to improve profitability, deposit interest rate was cut down sharply without a matching reduction of lending rate of interest, made possible due to the surge in retail lending (Mohan 2005, 2006).

Fiscal policy reforms have taken place through reduction in direct taxes as well as indirect taxes. Reduction of indirect taxes has been more rapid than direct income taxes. Illustratively, the peak rate of customs duty has come down sharply from as high as 130 per cent in 1992 to 12.5 per cent by 2006-07. However, effective personal income tax during the 1990s witnessed a reduction of only 0.70-1.80 percentage points as compared with the1980s for annual income in the range Rs.150 thousand to Rs 500 thousand, the income group covering largely the middle income class which account for a major share of household savings.

As regards the financial sector, capital market witnessed a surge during the 1990s, spurred by portfolio capital inflows from abroad. This is evident from the growth of stock market capitalisation, which has surpassed banking sector assets. However, the sharp accretion of wealth accompanied by buoyant secondary segment of the equity market has not translated to a significant growth of primary equity issues. There is no perceptible shift in household preference toward saving in equities.

A critical issue arises here as to whether above changes in monetary, fiscal and financial policies under reforms have significantly influenced India's gold demand. In what follows we seek answer to this question. In order to facilitate the empirical analysis, a brief discussion on methodology and stylised facts about gold demand follows.

IV. METHODOLOGY AND DATA

On the empirical plane, studies on gold demand can be grouped into two traditions. In the first tradition, which dominates the literature with studies on industrial countries, the movement of gold price is explained in terms of a set of macroeconomic and financial variables (Aggarwal and Soenen, 1988, Ghosh et al., 2002, Lawrence, 2003, Ranson and Wainwright, 2005, Capie et al., 2004, Mani and Vuyyuri, 2003). In the second tradition, analysts use physical gold demand as the dependent variable, explained by a set of macroeconomic variables and domestic relative price of gold (gold price relative to producer price or consumer price indices). Studies in this tradition are rather scarce and available in the Indian context (Patel, 1950, Sharma et al., 1992 and Vaidyanathan, 1999). At a theoretical level, studies in the first tradition are based on the equilibrium framework; demand and supply forces lead to a reduced form model of gold price, determined by major macroeconomic and financial factors (Ghosh et al., 2002, Lawrence, 2003). The equilibrium framework, however, works under the assumptions of perfect competition and efficient market hypothesis. Empirical studies show that the distribution of daffy variation in global gold prices (London fixing) significantly departs from the condition for market efficiency (Aggarwal and Soenen, 1988). Moreover, for a specific country, the gold market may be either supply driven due to sufficient domestic production or demand driven due to large import demand. Countries like India depending entirely on gold import are price-takers, depending upon London fixes of gold prices, implying for an exogenous impact of gold price on physical gold demand. The domestic gold price is determined by global gold price, exchange rate, transaction cost, import duties and some arbitrage component. Thus, it is appropriate to analyse gold demand in physical form, depending on domestic gold price, and various other macroeconomic and financial variables. Given the above insights, the physical demand function for gold (G) in general could be postulated using income (y), relative gold price (p), and a set of policy variables such as interest rate (r), equity price (s), exchange rate (e), taxes (t), and government expenditure (w):

[G.sub.t] = [alpha] + [[beta].sub.1][p.sub.t] + [[beta].sub.2][y.sub.t] + [[beta].sub.3][r.sub.t] + [[beta].sub.4][s.sub.t] + [[beta].sub.5][e.sub.t] + [p.sub.6.sup.t] + [[beta].sub.7]w + [[epsilon].sub.t] (1)

where [epsilon] refers to the error term, and 't', time period subscript. All variables are measured in natural logarithm scale in order to facilitate the analysis in terms of elasticity response of gold demand to its principal determinants. Additionally, the contemporaneous or a period lag of world gold supply or mining production may be included in the demand function to gauge the impact of favourable supply condition.

Estimation of equation (1) is not straightforward without examining the stochastic characteristics of the variables. If [epsilon] is stationary and the variables are non-stationary, Engle and Granger (1987) suggests that the error correction model is the correct approach to account for both short-run and long run components. For a multivariate model, however, some variables ([X.sub.j]) may turn out to be non-stationary and others ([X.sub.k]) stationary processes. In this regard, a selection of [X.sub.j] and [X.sub.k] could be based on unit root test such as Augmented-Dickey Fuller and Phillips-Perron Unit Root Tests. In this case, the long-run trajectory could be defined over integrated variables and the stationary variables enter the short-run model consistent with economic theory. The error correction model (ECM) subject to an identified long-run trajectory of the dependent variable gold demand (G) can be represented as follows:

[G.sub.t] = [alpha] + [[beta].sub.j][X.sub.j,t] (2)

and

[DELTA][G.sub.t] = c + [[summation].sup.m.sub.(j=1)][[beta].sub.j][DELTA][X.sub.j,t] + [[summation].sup.n.sub.(k=1)][[beta].sub.k][delta][X.sub.k,t] + [theta][[epsilon].sub.t-1] + [v.sub.t] (3)

Equation (2) identifies the long-run model and equation (3) is the error correction representation for gold demand, with variables [X.sub.j] and [X.sub.k], referring to non-stationary and stationary variables, respectively, [epsilon] the error correction term from cointegrating equation (2), v the white noise disturbance term and the symbol ([DELTA]) denoting the first difference operator. The parameter [theta] measures the speed of adjustment to long-run path.

As regards data, this study uses annual data on India's gold offtake, which were culled out from the Annual Reports on Gold Survey by Gold Field Mines Survey (GEMS), London. The Handbook of Statistics for the Indian Economy, published by the Reserve Bank of India provided data on macroeconomic aggregates and financial variables. The definition of variables used in the empirical model is shown in Annex.

V. STYLIZED FACTS

India's gold consumption stood at 820 tonnes in 2005, equivalent to US $15 billion. Relating to macroeconomic aggregates, gold imports accounted for about 2.4 per cent, 2.0 per cent, 5.0 per cent and 8.0 per cent of private consumption, private disposable income, gross domestic savings and net domestic savings, respectively. Over a long horizon during the period 1980-2005, aggregate gold consumption grew at a trend rate of 10.5 per cent, more than twice the global gold demand growth at 4.5 per cent (Chart 1).

[GRAPHIC 1 OMITTED]

Gold consumption witnessed sharp acceleration during the 1990s amidst liberalization of gold import policy, strong economic growth and favourable movements in gold prices. During 1997-98, gold demand reached a record high at 830 tonnes. Although gold demand slowed thereafter, it maintained an elevated level until 2001. In 2003, gold demand declined by 20 per cent, mainly due to the economic slowdown triggered by monsoon failure and the decline in agriculture income. Subsequently, the sharp acceleration in gold price did not deter growth of India's gold consumption, as it posted annual increase of 20.1 per cent in 2004 and 14 per cent in 2005, supported by buoyant economy and deepening of financial markets. Domestic price of gold recorded a trend increase of 5.2 per cent during 1981-2005. In real terms, relative price of gold showed a declining trend, albeit, moderately at rate of 2.0 per cent.

Component-wise, high cartage jewellery consumption accounts for the major share of India's total gold demand, followed by investment demand and others, including dental and industrial demand, and coins and medals (5). Though gold jewellery demand continues to remain dominant, its share in total gold demand has declined from 97 per cent during the 1980s to 84.5 per cent during the 1990s and 78 per cent during 2000-05. The share of investment demand has increased from 2.5 per cent during the 1980s to 11 per cent during the 1990s and further to 16 per cent in 2005. India emerged as the leading country with a share of 48 per cent of world investment demand for gold in 2005. The sharp pick-up in investment demand occurred due to gold banking and introduction of gold futures trading since 2003-04. In the Multi Commodity Exchange, gold futures trading captured 50 per cent of average daily turnover.

Gold has crucial implications for India's balance of payments. Gold imports in US dollar terms accounted for about 44 per cent of merchandise trade deficit during 1999-00 to 2005-06. For some years, in particular, 2000-01 and 2001-02, gold import accounted for as high as 69 per cent and 55 per cent of India's merchandise trade deficit. Assuming there were no gold imports, the current account balance (CAB) would have shown a different picture. Illustratively, for the years 2000-01 and 2005-06, CAB would have shown a surplus as against a deficit. On the other hand, in the absence of gold imports, the surplus on current account balance would have accentuated during 2001-02 to 2004-05.

Finally, as regards the importance of gold consumption relative to bank deposits (rural and semi-urban areas), aggregate bank deposits of rural and semi-urban areas expanded by an average of Rs. 414 billion per annum during 1999-2000 to 2005-06 as compared with gold consumption of Rs. 234 billion i.e. 60 per cent of aggregate gold demand averaging Rs 390 billion. Over long-horizon, India's gold off-take on cumulative basis stood at Rs 5 trillion during 1975-2005, as compared with bank deposits at all-India level outstanding at Rs 21 trillion and rural and semi-urban bank deposits outstanding at Rs 5.3 trillion during 2005-06. This suggests that financial assets such as bank deposits are inadequate to mobilise potential savings in rural and semi-urban areas, leading to increasing saving in gold (Chart 2).

[GRAPHIC 2 OMITTED]

VI. EMPIRICAL FINDINGS

The Augmented Dickey-Fuller Unit Root Test revealed that physical gold consumption, per capita real income, gold price, and equity price (Bombay Stock Exchange Index) are non-stationary and first order integrated processes, while real interest rate a stationary process (Table 1). Accordingly, the long-run trajectory of gold demand is defined over per capita real income, price and wealth (stock price) variables in line with standard consumption or saving function. Real interest rate, being a stationary series, is included in the error correction model. Since gold mining production grows only slowly over long-horizon, and various other above ground sources of supply such as central banks' sales, and supply of scraps supplement aggregate gold supply in the short-run to match aggregate demand, the variable for changes in supply condition was included in the error correction model. In addition, the error correction model included variables pertaining to personal income tax and per capita real government expenditure to account for economic and social security. For robustness, the ECM was also experimented with alternative measures of income and interest rate such as agricultural income and real deposit rate of interest. The empirical exercise was replicated in three alternative forms pertaining to aggregate gold demand, aggregate fabrication, and jewellery demand. In order to provide robustness to empirical findings, alternative regression equations were estimated as shown across Table 2 through 7.

1. Long Run Trajectory

Beginning with the long-run gold demand, the empirical model shows that real income (per capita terms), gold price and wealth effects are highly statistically significant (Table 2). Real gold price has inverse relationship with gold demand. Real per capita income and real equity wealth have positive influence on gold demand. The coefficient of income, gold price and equity price are more or less similar across aggregate gold demand and fabrication and jewellery demands. The long-run models have high coefficient of determination (adjusted [R.sup.2]) and they are statistically sound as the DW statistics is higher than the adjusted [R.sup.2]. The error terms associated with the long-run models turned stationary in terms of Cointegrating Regression Durbin Watson (CRDW) test.

2. Error Correction Model

The error correction model shows some common features across the three demand functions for gold (Table 2). A major strength of the error correction model is the high adjusted [R.sup.2] despite the variables in first difference form, suggesting that the fitted values track the actual values of the dependent variable. The adjusted [R.sup.2] and DW statistics are almost similar for aggregate gold demand and fabrication and jewellery demands. The error correction term is negative and highly significant. In terms of coefficient size, the error correction term is almost similar across alternative demand functions. Most importantly, the size of error correction-term indicated rapid adjustment of gold demand to deviations from the long-run path. Despite these common features, empirical gold demand shows some notable features in response to explanatory variables as discussed in the following.

3. Impact of Key Variables

First, India's gold demand is significantly price elastic, in the short-run as well as longer horizon. Relative price of gold has inverse relationship with gold demand. At the same time, gold demand is significantly responsive to income, wealth, risk free real interest rate, global supply conditions, personal income taxes and government capital expenditure.

Second, a key feature of long-run gold demand is that in terms of coefficient size, the price elasticity is substantially higher than income elasticity and wealth effect (Table 2). Illustratively, for the aggregate gold demand the long run price elasticity of aggregate gold demand (1.81) is about twice larger than the income elasticity (0.74) and wealth effect (0.83). It is evident that for the aggregate gold demand the sum of long run coefficients of income and wealth turns closer to price elasticity coefficient in the long run. This implies that for long run growth of gold demand, income and wealth should significantly outpace real gold price. It was noted earlier that relative price of gold remained more or less stable, albeit a modest declining trend over longer horizon. Thus, despite high price elasticity, the sources of long run growth in India's gold demand could be attributable to income and wealth effect.

Third, real income has positive and asymmetric effect on gold demand with different short-run and long run effects. Income elasticity of gold demand is less than unity in the longer run but close to unity in the short run. Income elasticity could be about half of the price elasticity of gold demand. Additionally, there is some evidence that gold jewellery demand may be more elastic to rural income than aggregate income in the longer run.

Fourth, financial wealth, induced by medium term trends in real equity prices, has higher positive impact on gold demand in the long run than the short-run. Equity price elasticity of gold demand is much lower than relative gold price and income effects. However, since equity prices are prone to sharp fluctuations, at times, their actual impact could be much higher than income and gold price effect.

Fifth, real yield on government bonds have inverse relationship with gold demand. Thus, gold serves as an alternative instrument for households saving. In the presence of agriculture income, interest elasticity turns lower, implying that gold jewellery demand in rural areas may not be as sensitive to interest rate as in urban areas.

Sixth, a surprising result was that real deposit rate of interest could not have statistically significant effect on gold demand (Table 3). This could be attributable to low attractions of rural areas to bank deposits and the lack of flexibility in deposit rate of interest. Financial innovation, in the form of development of various segment of the financial sector, such as mutual funds, equity market etc., which have implications for urban gold demand, could be another factor.

Seventh, exchange rate has an inverse relationship through gold price in domestic currency. Nominal exchange rate variation indicating asset demand effect could not be robust in the presence of other variables such as taxes and government expenditure (Table 4 and Table 7).

Eighth, favourable supply conditions have a positive effect on India's gold demand, attesting to scarcity value of gold.

Ninth, a statistically significant positive impact of marginal tax rate on personal income on gold jewellery demand could emerge for the differential between average tax rate pertaining to an income of Rs. 0.5 million and the tax rate on income level of Rs 0.2 million (Table 5). The tax variable did not adversely affect the sign condition, coefficient size and statistical significance of key variables such as income, gold price and equity prices.

Tenth, higher government capital expenditure relative to aggregate expenditure reduces gold demand for precautionary saving purposes (Table 6). A numerical simulation in this regard revealed that an increase in the share of government capital expenditure by 5-percentage point from the current level of about 15 per cent to 20 per cent could have significant effect on stabilising gold demand. Empirical findings also suggest that in the presence of taxes and government capital expenditure, short-run income effect on gold demand could be moderated and remain closer to unity. The income elasticity of jewellery demand with respect to agricultural income in particular could witness a significant moderation in the presence of taxes and government capital expenditure (Table 7).

Finally, as regards other variables, it was found that domestic prices of silver (real terms) could not have significant impact on gold demand. A similar finding was with respect to the margin variable characterising the difference between domestic and international gold prices. A critical finding was that though silver price and margin variables were statistically insignificant, their presence could cause statistical insignificance of real income impact on gold demand. This finding in particular suggests statistical problems of using correlated variables, such as price of substitutes, leading to alternative perspectives of income effect on gold demand.

4. Model Stability

Empirical models need to be checked for stability in order to exploit them for policy analysis. In this regard, cumulative sum of recursive residuals (CUSUM) and cumulative sum of squares of recursive residuals (CUSUMSQ) test showed overall stability of the ECM. Illustratively, the CUSUM and CUSUM square tests for aggregate gold demand (Equation 21) showed that residuals of the ECM remained within two standard error bands during the period 1991-2005. Furthermore, structural stability of the principal component of gold demand, i.e. aggregate jewellery demand (equation 23) during the period of the 1990s was investigated using Chow Break Point test. The resulting F test showed that jewellery demand exhibited stability during 1991-1997. For the remaining period, stability in jewellery demand was evident from Chow's forecast test (Table 8).

VII. POLICY PERSPECTIVE

It is possible to derive some useful policy perspectives based on our empirical findings. Broadly, policy perspectives could relate to economic development goals in the context of 'inclusive growth' and 'financial inclusion' on the one hand and development of gold market on the other. First, the relationship of gold demand with real income, inflation, and asset prices has implications for policy goals pertaining to economic growth, price stability and financial stability. Gold demand in the short-run can be modulated with policy actions through interest rate, exchange rate, taxes and government spending, which constitute monetary and fiscal instruments. Second, while buoyant gold demand suggests that there is substantial potential for augmenting savings for productive purposes, it also reflects continuing rigidities in the financial system and the problem of inadequate access of people to formal sector such as banks. Recognising this, monetary, fiscal and financial sector polices need aim at improving access of people to financial system while providing adequate incentives for financial savings in terms of three parameters, safety, return, and liquidity. For rural areas, bank deposits and postal deposits are the avenues for financial savings. However, bank deposits lack attractiveness in terms of real return. Thus, monetary and financial policy initiatives through interest rate, exchange rate and asset prices have to contend with the challenges of enticing banks to provide adequate real interest on retail deposits, particularly in rural areas. Liquidity of financial instruments is a major concern, since gold is a highly liquid instrument. Gold's role in providing spot insurance against contingencies cannot be matched with any financial instrument. Liquidity of financial instruments can be enhanced by way of information technology enabled services of banks and innovative payment mechanism, but these are by and large concentrated in urban and metropolitan regions. Thus, a need is felt for policy initiatives toward increasing the outreach of technology enabled services from banks to rural areas. Third, there are two major lessons for fiscal policy. One, the significant response of gold demand to personal income tax suggests that besides 'safe haven', gold also provides 'tax haven' for middle class households having greater propensity of saving. Thus, a moderate and effective personal income tax regime may play an important role in the financial intermediation process and the management of gold demand. Two, economic and social uncertainties, which fuel gold demand despite robust economic growth, could be ameliorated through credible fiscal policy and enhanced government spending on provisions of social security and physical infrastructure. Fourth, policies toward better financial intermediation need not be construed as strategies for weaning away people from gold. A liberal gold policy, leading to better pricing of gold in the domestic market, would be beneficial for the consumers and the society as a whole. Price discovery through an efficient gold market would enable people to evaluate alternatives such as financial saving. At present, domestic gold price is largely determined by global price and exchange rate, with arbitrage component vanishing due to liberalisation of gold trade. Thus, a two-pronged policy approach is called for. One, despite being the largest consumer, India is a price taker. Efforts should be made toward India's active role in price discovery in international gold market. Two, exchange rate policy need to assign some space for manoeuvring gold demand, since it is the second largest component of import.

Finally, going beyond findings of the study, there are some systemic issues relating to social and cultural environment, which play a vital role in influencing the access of people to formal financial sector and preference for gold savings. It is to be recognised that a rural worker can buy a small piece of gold jewellery from a local jewellery shop. However, it may be impossible for him to open a small deposit account in a bank, which insists on various documentary proofs of residence, income and introduction by an existing account holder. There are also various problems with credit facilities extended by banks. In the competitive environment, banks have to contend with the transaction cost associated with servicing retail deposits and credit accounts. Information technology could play an important role in facilitating retail banking in rural areas. However, it is perceived that literacy rate in rural areas is a major obstacle for enhancing the outreach of IT enabled banking services. In this context, there is recognition for the role of fiscal empowerment relating to spending on social sector such as education. However, the difference in literacy rate between rural and urban areas is not sharp enough to eschew the disproportionate provision of modern amenities for financial services in the two areas. Moreover, in the Indian context, gold will always have an edge over financial instruments in respect of convenience, transaction cost, and insurance against unforeseen events. However, complacency is uncalled for and efforts should be made toward enhancing the access of people to services from formal financial sector in order step up productive savings in the economy.

VIII. CONCLUSION

This study provided analytical and empirical perspectives on India's physical gold demand during the period 1980-2005. India's gold demand function was estimated in respect of aggregate gold demand as well as for fabrication and jewellery components. The demand function was specified in terms of real income, relative gold price and a set of variables such as interest rate, equity price, exchange rate, personal income tax, and government spending, which have a connect with monetary, fiscal and financial sector policies. Additionally, the impact of favourable global gold supply was underpinned. Empirical results revealed that gold demand is not only price sensitive, but also significantly influenced by macroeconomic and financial variables. The response of gold demand to these variables reveals as to how gold serves as a saving instrument in the Indian context. Based on empirical results, policy perspectives were derived in the context of economic development goals. From the perspective of further research, there is a need to address the issue whether gold constitutes a productive asset in India. Such a research will have far reaching implications for policies relating to gold as well as the macroeconomic environment. Given the sample size of the study, a cross-section study may provide further insights on the subject from a generalised perspective. Annex: Definition of Variables Variables Description Gold Demand (G) Natural logarithm of Aggregate Gold demand (tonnes) Gold Demand ([G.sub.F]) Natural logarithm of Aggregate Fabrication Gold demand (tonnes) Gold Demand ([G.sub.J]) Natural logarithm of Aggregate Jewellery demand (tonnes) including scrap demand Real Per capita Income (y) Natural Logarithm of Per capita Real Gross Domestic Product, i.e., GDP at factor cost at constant prices Real Gold Price (p) Natural logarithm of Domestic price of gold in Rupees deflated by wholesale price index Real Equity Price (s) Natural Logarithm of Three-year moving equity prices (Bombay Stock Exchange Index-BSE Sensex) deflated by wholesale price index Real Interest Rate (r) Natural logarithm of Interest rate on medium-term Government bonds adjusted by inflation rate: In [(1+ R) / (1+ p)1 Real Deposit interest Natural logarithm of Interest rate on rate (rd) medium-term bank deposits (3-5 years) adjusted by inflation rate: In [(1+ [R.sub.D]) / (1+ p)1 Supply Condition (Gs) One period lag of global gold supply defined over natural logarithm scale Tax Rate ([t.sub.k]) Natural logarithm of maximum income tax rate for income group of Rs 500,000 relative to income tax rate for income group Rs 200,000 Government Capital Natural Logarithm of Share of Government Capital Expenditure (W) Expenditure in Total expenditure Agricultural Income Natural Logarithm of Real GDP originating from agriculture sector Exchange Rate (e) Natural Logarithm of nominal exchange rate: Rupees per US dollar [DELTA] difference operator

The views expressed in the paper are solely of the authors and not of the institution to which they are attached. We are grateful to Dr. T. K. Chakrabarty and Dr. M.D. Patra for valuable suggestions. Contact address: ramalingakannan@rbi.org.in scdhal@rbi.org.in

References

Abken, P. A. (1980), "The Economics of Gold Price Movements", Commodity Journal, Vol. 14, Issue 4.

Abrams, O. (2005), 'Supply, not Demand, to Impact long-term Gold Price', Platt's Metals Week, Vol. 76, Issue. 18.

Adrienne, R. (2003), "Speculators Put Metal Analysts on Gold Watch", Financial Times, London, January.

Aggarwal, R. and Soenen, L. A. (1988), 'The Nature And Efficiency of Gold Market', Journal of Portfolio Management, Spring, Vol. 14, No. 3.

Baker, Stephen A., Van Tassel, Roger C. (1985), "Edwardsville: Forecasting the Price of Gold: A Fundamentalist Approach", Atlantic Economic Journal, December 1985. Vol. 13, Issue. 4.

Barsky, R. B. and Summers, L. H. (1985), "Gibson's Paradox and the Gold Standard", NBER Working Paper, No. 1680, National Bureau of Economic Research, US.

Bhattacharya, H. (2002), "Deregulation of Gold In India", World Gold Council, Research Study No 27, September.

Bream, R. (2004), "Sharp Rise in Gold Price Fails to Put off Buyers", Financial Times, London, June 4.

Cai, J., Cheung, Y. and Wong, M. (2001), "What Moves the Gold Market?" The Journal of Futures Markets, Vol. 21, No. 3.

Caldwell, J. Alexander (1980), "Gold: The Fundamentals Behind the Frenzy", Euromoney, London: February 1980.

Capie, F., Mills, T. C. and Wood, G. (2004), "Gold As A Hedge Against The Us Dollar", World Gold Council, Research Study No. 30, September.

Chandravarkar, A. (2006), "Introductory Essay on I G Paters Study on India's Elasticity of Demand for Gold, Economic and Political Weekly, February 2006.

Chua, J. and Woodward, R. (1982), "Gold as an Inflation Hedge: A Comparative Study of Six Major Industrial Countries", Journal of Business Finance and Accounting, Vol. 9.

Claudia, P. (1984), "Gold: Who Wants the Stuff?" Euromoney, London: December 1984.

Dalglish, B. (1993), "Going for the Gold", Maclean's Review, Toronto, June 7, 1993. Vol. 106, Issue. 23.

Engle, R. F. and Granger, C. W. J. (1987), "Cointegration, and Error Correction: Representation, Estimation, and Testing", Econometrica, Vol. 55, p. 251-76.

Feldstein, M. (1980), "The Effects of Inflation on the Prices of Land and Gold", Journal of Public Economics, Vol. 14.

Georgio, E. M. de. (1997), "What Affects the Price of Gold?r, African Business, April, Issue 220.

Ghosh, D., Levin, E. J. and Macmillan, P. (2002), "Gold as an Inflation Hedge?" Studies in Economics and Finance, Vol. 22, No. 1.

Gold Field Mines Survey (GFMS), U. K., Annual Survey on Gold (various reports).

Government of India (2006), 'Draft Approach to 11th Five Year Plan', Planning Commission.

Harmston, S. (1988), "Gold as a Store of Value", World Gold Council, Research Study No. 22, November.

Jaffe, J. F. (1989), "Gold and Gold Stocks as Investments for Institutional Portfolio", Financial Analysis Journal, March/April.

Judge, G. G., Griffiths, W. E., Hill, R. C. and Lutkepold, H. (1985), "Theory and Practice of Econometrics", John Wiley, New York.

Kavalis, N. (2006), "Commodity Prices and the Influence of the US Dollar", A Report Published by World Gold Council and GFMS Ltd, London, January.

Kettle, P. (2005), "The Size and Structure of Metals Markets: How Gold Compares with other Non-Ferrous Metals", World Gold Council Report, August.

Klapwijk, Philip (2003), "Medium-to Long-Term Outlook for Gold and Silver", The LBMA Precious Metals Conference 2003, Lisbon.

Koutsoyiannis, A. (1983), "Short-run Pricing Model for a Speculative Asset, Tested with Data from the Gold Bullion Market", Applied Economics, Vol. 15.

Lawrence, Colin (2003), "Why is Gold Different from other Assets?" An Empirical Investigation, World Gold Council, Publications, March 2003.

Levin, E. J. and Wright, R. E. (2006), "Short-run and Long-run Determination of the Price of Gold", World Gold Council. Research Paper No. 32, June.

Mani, Ganesh and Srivyal Vuyyuri (2003), "Gold Pricing in India: An Econometric Analysis", Economic Research, Vol. 16, No. 1.

Mohan, Rakesh (2005), "Inclusive Growth and Financial Inclusion", Reserve Bank of India, Speeches, Monthly Bulletin.

Parks, Lawrence (2000), "The Near Death & Resurrection of the Gold Mining Industry", July 17. (available from Google Search on world wide web).

Patel, I. G. (1950), "India's Elasticity of Demand for Gold, August, International Monetary Fund", Research paper 27. Reprinted in Economic and Political Weekly, February 2006.

Pulvermacher, K. (2004), "Gold and Hedge Funds: A Comparative Analysis", World Gold Council Report, June.

Pulvermacher, K. (2004), "United States And United Kingdom: Analysis of Long-Run Correlation of Returns On Gold and Equities", World Gold Council, January.

Pulvermacher, K. (2005), "Commodity Returns and The Economic Cycle", World Gold Council Report, May.

Ransen, D. and H. C. Wainwright (2005), "Why Gold, Not Oil? Is the Superior Predictor of Inflation", World Gold Council Research, November.

Rao, B. S. and Nagabhushanam, K. (1960), "India's Demand for Import of Non-monetary Gold, non-monetary Silver and Merchandise, 1901-1913", Unpublished paper.

Reddy, Y., V. (1997), "Gold Banking in India", BIS Review, No.76/1997, Bank for International Settlements, Switzerland.

Reddy, Y., V. (2002), "Evolving Role of Gold--Recent Trends and Future Direction", BIS Review, No. 21-2002, Bank for International Settlements.

Rhona O'Connell (2003), "India: Exciting Advances in the Local Gold Market", World Gold Council Research.

Sharma, A., Vasudevan, A., Sabapathy, K. and Roy, M. (1992), "Gold Mobilisation Instrument as an External Adjustment", Reserve Bank of India Discussion Paper, DRG Studies, RBI.

Sherman, E. J. (1983), "A Gold Pricing Model", Journal of Portfolio Management, New York: Spring 1983. Vol. 9, Issue. 3.

Shirvani, H. and Wilbratte, B. (2000), "Does Consumption Respond More Strongly to Stock Market Declines than to Increases?", International Economic Journal, Volume 14, Number 3, Autumn 2000.

Sjaastad, L. and Scacciavillani, F., (1996), "The Price of Gold and the Exchange Rate", Journal of International Money and Finance, Vol. 15, No. 6.

Smets, Frank (1997), "Financial Asset Priecs and Monetary Policy, Theory and Evidence", BIS Working Papers, No. 47, Bank for International Settlements, Basle, Switzerland.

Smith Graham (2001), "The Price of Gold and Stock Market Indices for the United States", Unpublished Paper, November, 2001. http://www.lyxorgbs.it/pdf/gold_usstockindicesdec200120.pdf

Thorat, Usha (1997), "Development of Gold Market in India", Lecture at the World Gold Council Conference, New Delhi, on Saturday, August 2, 1997.

Troost, B. W. (2006), "Analyzing the Determinants of the Gold Price", June. mimeo. (Google Search).

Vaidyanathan, A. (1999), "Consumption of Gold in India", Economic and Political Weekly, February.

World Gold Council: The Value of GOld to Society, http://www.gold.org.

R. KANNAN & SARAT DHAL

Department of Economic Analysis & Policy, Reserve Bank of India

Notes

(1.) See the Draft Approach to 11th Five Year Plan for a discussion on 'Inclusive Growth' as major objective of Government's macroeconomic policy in India. According to the 11th Five Year Plan document, despite rapid economic progress, India continues to remain as the abode for largest number of poor people. Though poverty rate has come down, the process has been rather slow. A sizeable number of people are yet to experience an improvement in standard of living, in terms of safe drinking water, education, health and sanitation, electricity consumption, and access to financial services.

(2.) Given the sharp volatility in equity prices in the short-run, it is appropriate to capture the equity price induced wealth effect by using a measure of underlying equity prices, such as the geometric mean of equity price over a medium term horizon (three to five years).

(3.) Reform of the banking and financial sector and liberalisation of external sector based on the recommendations of the Committee on Banking Sector Reforms (Chairman: M. Narasimham) and the High level Committee on balance of Payment (Chairman: Dr. C. Rangarajan).

(4.) The high caratage gold jewellery demand makes India different from advanced countries. This reflects mainly the saving motive in India as compared with consumption of fashion jewellery in industrial countries. Table 1 ADF Unit Root Test Variables Variables in Levels (Log scale) With Intercept (C) With C & Trend Gold Demand (GIND) GIND-Total -2.28 -2.59 GIND-Fabrication -2.42 -2.58 GIND-Jewellery -2.44 -2.72 Real Gold Price -2.25 -2.28 Equity Price (BSE Sensex) -0.96 -1.68 Real Interest Rates Up to 3-yr Deposits (DR3) -3.09 3-5 yr Deposits -2.83 Yield on Medium-Longer -3.83 maturity Government Bonds Lending rate of interest (LR) -3.79 Variables Variables in First Difference With C Gold Demand (GIND) GIND-Total -3.92 GIND-Fabrication -3.97 GIND-Jewellery -4.04 Real Gold Price -4.95 Equity Price (BSE Sensex) -4.99 Real Interest Rates Up to 3-yr Deposits (DR3) 3-5 yr Deposits Yield on Medium-Longer maturity Government Bonds Lending rate of interest (LR) * MacKinnon (1996): critical values for ADF model with intercept term are 2.99 and 2.61 at 5% and 10% level of significance. For models with intercept and trend terms, the critical values are 3.61 and 3.24 for 5% and 10% level of significance, respectively. Table 2 Estimated Gold Demand in India Equation 1 Equation 2 Equation 3 Aggregate Fabrication Jewellery Demand Demand Demand Long Run Intercept(?? 10.0279 9.8482 10.444 (5.09) (5.00) (5.14) Real Per-capita Income (y) 0.7419 0.7381 0.6685 (5.15) (5.16) (4.40) Real Gold Price (p) -1.8086 -1.7947 -1.8239 (-7.3) (-7.60) (-7.09) Real Equity Price (s) 0.829 0.7614 0.7455 (10.59) (9.89) (8.90) [R.sup.2] 0.9845 0.9844 0.9808 DW 1.431 1.300 1.17 Error Correction Model Intercept ([beta]o) 0.0707 0.0502 0.0487 (2.32) (1.91) (1.83) Real Per capita Income 1.108 1.1931 1.2409 ([DELTA]y) (2.59) (3.04) (3.04) Real Gold Price ([DELTA]p) -2.0808 -1.9549 -1.9978 (-15.39) (-16.48) (-19.43) Real Equity Price ([DELTA]s) 0.4531 0.3896 0.3812 (3.73) (3.82) (3.63) Real Interest Rate (r) -2.2646 -1.7618 -1.8666 (-2.63) (-2.37) (-2.44) Supply Condition ([DELTA]Gs) 0.3839 0.5019 0.4968 (2.28) (3.75) (3.56) Error Correction -0.7100 -0.7475 -0.6896 ([[epsilon].sub.1]) (-4.71) (-6.25) (-6.18) Adjusted [R.sup.2] 0.8978 0.9089 0.9069 DW 1.92 1.91 1.91 Equation 4 Equation 5 Jewellery Jewellery Demand Demand (Agriculture (Real GDP) Income) Long Run Intercept(?? -0.2373 6.768 (-0.06) (2.88) Real Per-capita Income (y) 1.1841 0.539 (4.72) (4.90) Real Gold Price (p) -1.7716 -1.644 (-6.64) (-7.02) Real Equity Price (s) 0.6411 0.6520 (6.67) (8.13) [R.sup.2] 0.982 0.979 DW 1.40 1.210 Error Correction Model Intercept ([beta]o) 0.0761 0.0235 (3.43) (0.84) Real Per capita Income 0.7953 1.2496 ([DELTA]y) (2.55) (3.07) Real Gold Price ([DELTA]p) -1.9196 -2.0027 (-18.13) (-19.55) Real Equity Price ([DELTA]s) 0.3309 0.3740 (2.65) (3.54) Real Interest Rate (r) -1.7500 -1.8144 (-2.36) (-2.38) Supply Condition ([DELTA]Gs) 0.4281 0.4841 (2.75) (3.49) Error Correction -0.730297 -0.6873 ([[epsilon].sub.-1]) (-4.98) (-5.95) Adjusted [R.sup.2] 0.8933 0.9069 DW 1.760 1.930 Figures in parentheses indicate 't' statistics. The critical 't' values for 5 per cent and 10 per cent level of significance are about 2.0 and 1.80, respectively. Table 3 Estimated Gold Demand in India (with Deposit Rate of Interest) Equation 6 Equation 7 Equation 8 Aggregate Fabrication Jewellery Gold Demand Demand Demand Intercept ([beta]o) 0.0487 0.0408 0.03679 (0.87) (0.82) (0.71) Real Per capita Income 0.8748 0.9717 1.0065 ([DELTA]y) (1.40) (1.87) (1.90) Real Gold Price ([DELTA]p) -2.123 -1.9953 -2.0346 (-12.41) (-13.65) (-15.26) Real Equity Price ([DELTA]s) 0.5073 0.4162 0.4126 (3.93) (3.28) (3.09) Real Interest Rate (rd) -1.0484 -0.9856 -0.9889 (-0.00) (-0.00) (-0.00) Supply Condition ([DELTA]Gs) 0..3719 0.4933 0.488 (1.52) (2.75) (2.67) Error Correction -0.7793 -0.7842 -0.7255 ([[epsilon].sub.-1]) (-5.12) (-5.80) (-5.50) Adjusted [R.sup.2] 0.8617 0.8855 0.8807 DW 1.56 1.57 1.54 Equation 9 Equation 10 Jewellery Jewellery Demand Demand (Agr. Income) (Real GDP) Intercept ([beta]o) 0.0530 0.0126 (1.82) (0.21) Real Per capita Income 0.8046 1.0484 ([DELTA]y) (2.71) (1.88) Real Gold Price ([DELTA]p) -1.9477 -2.0350 (-15.29) (-15.32) Real Equity Price ([DELTA]s) 0.3691 0.4095 (2.70) (3.09) Real Interest Rate (rd) -0.8593 -0.9016 (-0.00) (-0.00) Supply Condition ([DELTA]Gs) 0.4294 0.4774 (2.48) (2.67) Error Correction -0.7678 -0.7241 ([[epsilon].sub.-1]) (-4.90) (-5.34) Adjusted [R.sup.2] 0.8687 0.8809 DW 1.530 1.570 Figures in parentheses indicate 't' statistics. The critical 't' values for 5 per cent and 10 per cent level of significance are about 2.0 and 1.80, respectively. Table 4 Estimated Gold Demand in India (with exchange rate) Equation 11 Equation 12 Equation 13 Aggregate Fabrication Jewellery Gold Demand Demand Demand Intercept ([beta]o) 0.0212 0.0056 0.0117 (0.57) (0.22) (0.40) Real Per capita Income 1.7530 1.8103 1.7489 ([DELTA]y) (2.86) (3.68) (3.32) Real Gold Price ([DELTA]p) -2.1149 -1.9900 -2.0266 (-15.30) (-17.91) (-17.95) Real Equity Price ([DELTA]s) 0.4570 0.3856 0.3769 (4.56) (4.71) (4.01) Real Interest Rate (r) -2.1625 -1.7333 -1.8429 (-2.78) (-2.76) (-2.73) Supply Condition ([DELTA]Gs) 0.2969 0.4126 0.4287 (1.88) (3.01) (2.81) Exchange Rate ([DELTA]e) 0.3884 0.3723 0.3088 (1.97) (2.14) (1.68) Error Correction -0.8100 -0.8475 -0.7737 ([[epsilon].sub.-1]) (-4.66) (-5.37) (-5.21) Adjusted [R.sup.2] 0.9107 0.9229 0.9155 DW 1.99 2.10 2.05 Figures in parentheses indicate 't' statistics. The critical 't' values for 5 per cent and 10 per cent level of significance are about 2.0 and 1.80, respectively. Table 5 Estimated Gold Demand in India (with marginal income tax) Equation 14 Equation 15 Equation 16 Aggregate Fabrication Jewellery Gold Demand Demand Demand Intercept ([beta]o) 0.0194 0.0033 0.0092 (0.54) (0.13) (0.30) Real Per capita Income 1.6727 1.7270 1.6656 ([DELTA]y) (2.85) (3.50) (3.07) Real Gold Price ([DELTA]p) -2.1074 -1.9818 -2.019 (-15.09) (-18.01) (-18.21) Real Equity Price ([DELTA]s) 0.4604 0.3902 0.3819 (5.05) (5.81) (4.80) Real Interest Rate (r) -2.0112 -1.5665 -1.6701 (-2.65) (-2.56) (-2.50) Supply Condition ([DELTA]Gs) 0.3388 0.4571 0.4740 (2.28) (3.62) (3.39) Exchange Rate ([DELTA]e) 0.3626 0.3455 0.2812 (2.04) (2.25) (1.70) Tax Rate ([DELTA][t.sub.k]) 0.0595 0.0622 0.0641 (1.93) (2.06) (2.41) Error Correction -0.7990 -0.8357 -0.7637 ([[epsilon].sub.-1]) (-4.76) (-5.65) (-5.53) Adjusted [R.sup.2] 0.9161 0.88 0.9226 DW 2.16 2.20 2.240 Figures in parentheses indicate 't' statistics. The critical 't' values for 5 per cent and 10 per cent level of significance are about 2.0 and 1.80, respectively. Table 6 Estimated Gold Demand in India (with Government Expenditure) Equation 20 Jewellery Equation 18 Equation 19 Demand Aggregate Jewellery (Agriculture Gold Demand Demand Income) Intercept (([beta]o) 0.0514 0.0367 0.0873 (1.13) (1.06) (3.42) Real Per capita Income 1.2583 1.3047 0.4275 ([DELTA]y) (2.17) (2.40) (2.25) Real Gold Price ([DELTA]p) -1.9932 -1.9171 -1.8776 (-14.21) (-15.15) (-14.90) Real Equity Price ([DELTA]s) 0.3794 0.3095 0.2719 (3.44) (3.46) (3.00) Real Interest Rate (r) -2.1779 -1.8004 -1.7288 (-2.71) (-2.69) (-2.68) Supply Condition ([DELTA]Gs) 0.3532 0.4869 0.4703 (2.36) (3.63) (3.58) Exchange Rate ([DELTA]e) 0.1496 0.09 -0.1050 (0.89) (0.64) (-1.06) Tax Rate ([DELTA][t.sub.k]) 0.0930 0.0951 0.1022 (3.92) (3.57) (4.53) Government Capital -0.2976 -0.2787 -0.3293 Expenditure ([DELTA]W) (-2.11) (-1.88) (-2.35) Error Correction -0.7705 -0.7522 -0.7140 ([epsilon].sub.-1]) (-6.01) (-6.44) (-6.47) Adjusted [R.sup.2] 0.9387 0.9448 0.9434 DW 1.65 1.97 1.94 Figures in parentheses indicate 't' statistics. The critical 't' values for 5 per cent and 10 per cent level of significance are about 2.0 and 1.80, respectively. Table 7 Estimated Gold Demand in India (Final Model) Equation 21 Equation 22 Equation 23 Aggregate Fabrication Jewellery Gold Demand Demand Demand Intercept (([beta]o) 0.0707 0.0490 0.0475 (2.68) (2.12) (2.05) Real Per capita Income 1.0018 1.1059 1.1542 ([DELTA]y) (2.81) (2.91) (2.96) Real Gold Price ([DELTA]p) -1.9701 -1.8671 -1.9036 (-16.20) (-17.40) (-16.88) Real Equity Price ([DELTA]s) 0.3700 0.3223 0.3063 (3.23) (3.25) (3.35) Real Interest Rate (r) -2.2203 -1.6952 -1.8094 (-2.90) (-2.50) (-2.78) Supply Condition ([DELTA]Gs) 0.3852 0.5113 0.5059 (2.55) (3.89) (4.10) Tax Rate ([DELTA][t.sub.k]) 0.0995 0.0959 0.0988 (4.17) (3.82) (4.15) Government Capital -0.3278 -0.2667 -0.2956 Expenditure ([DELTA]W) (-2.48) (-1.90) (-2.15) Error Correction -0.7344 -0.7763 -0.7304 ([epsilon].sub.-1]) (-6.43) (-7.22) (-6.99) Adjusted [R.sup.2] 0.9371 0.9429 0.9442 DW 1.60 1.91 1.920 Equation 24 Equation 25 Jewellery Jewellery Demand Demand (Agr. Income) (Real GDP) Intercept (([beta]o) 0.0719 0.0256 (3.19) (0.94) Real Per capita Income 0.8410 1.1294 ([DELTA]y) (3.42) (2.84) Real Gold Price ([DELTA]p) -1.8166 -1.9088 (-11.87) (-16.87) Real Equity Price ([DELTA]s) 0.2471 0.2974 (3.18) (3.43) Real Interest Rate (r) -1.7429 -1.7492 (-2.69) (-2.72) Supply Condition ([DELTA]Gs) 0.4318 0.4935 (3.81) (4.08) Tax Rate ([DELTA][t.sub.k]) 0.0892 0.1006 (3.89) (4.24) Government Capital -0.3263 -0.3019 Expenditure ([DELTA]W) (-2.41) (-2.26) Error Correction -0.7912 -0.7348 ([epsilon].sub.-1]) (-5.65) (-6.90) Adjusted [R.sup.2] 0.9341 0.9455 DW 1.66 1.91 Figures in parentheses indicate 't' statistics. The critical 't' values for 5 per cent and 10 per cent level of significance are about 2.0 and 1.80, respectively. Table 8 Structural Break and Stability of Gold Jewellery Demand Chow Break Point Tests (Years) F statistics (probability) Null Hypothesis: No Structural Break 1991 0.78 (0.65) 1992 0.77 (0.65) 1993 0.88 (0.59) 1994 0.87 (0.59) 1995 0.77 (0.65) 1996 0.38 (0.91) 1997 1.95 (0.22) 1998-2005 (Chow Forecast Test) 0.71 (0.68)