7. ECONOMETRIC TOOLS IN UNDERSTANDING THE RELATIONSHIP BETWEEN MONEY, OUTPUT, AND PRICES

Studies like in late 1980s by Ahmed, Nachane etc reported that: 2.

1. A series of papers started coming up since 1973 on the relationship among money, output and prices in the Indian economy though Bhattacharya (1972) estimated the relationship between money and output in India with the simple OLS technique; till 1982 in India, research works in this area covered unidirectional causations like from money supply to output and from money supply to prices; after Granger (1969) was published, attention of researchers went to­wards causality between a pair of economic 3.

time series among forecasters in studies of the causal effects of one series on the other;

Sims (1972) undertook an exercise based on Granger to detect casual direction between money and income in the post-war data; from the analysis Sims observed that while income did not cause money, money caused income; on the same principle a more detailed study was undertaken by Pierce (1977) to establish the causal relationships between several pairs of variables in the US economy; he observed that numerous economic variables which were generally regarded as strongly interrelated might, with equal validity based on recent empirical evidence, be regarded as independent or weakly related; on the Indian data, Nachane and Nadkarni (1985)

undertook a similar study and found that high powered money caused broad money M₃ for short period, M₃ affected price in long run; Sims (1972) in addition to Granger (1969) influenced the studies on Indian economy involving causality; Sims (1972) applied the Granger causality to test for evidence of unidirectional causality between money and nominal income in the U.S. for the period 1947-69; since 1983 papers on causality be­tween money supply and output and between money supply and prices started coming up in India.

Brunner and Meltzer (1964) and Cagan (1965) created of curiosity among the Indian researchers on money multiplier ‘m’ in the identity M = mR, where M was money sup­ply aggregate, R was reserve money; two issues of particular concern, stability and predictability were crucial determinants of the superiority or otherwise of the monetary base as an instrument of money-stock control vis-a-vis interest rates as per Pierce (1977) because there developed an extensive litera­ture on money-multiplier forecasting models revolving around these two issues.

Burger, Kalish and Babb (1971) envisaged the monetary policy problem facing the US Federal Reserve Board, as one of determin­ing the optimal level of the monetary base for a targeted level of money supply in the face of a stochastically fluctuating money­multiplier; the multiplier was modelled by a regression analysis involving its own lagged values and certain other economic variables; Bomhoff (1977) took a somewhat direct view of the problem and modelled ‘m’ as a univariate Box-Jenkins ARIMA process; his approach was refined and elaborated in a series of papers by Johannes and Rasche (1979, 1981, 1982) who formulated ARIMA models for the components of ‘m’ (such as the currency-deposit ratio, bank borrowing ratio, adjusted reserve ratio etc) on the pre-

supposition that the components approach being based on more disaggregated informa­tion Hafer and Hein (1984) contradicted this supposition with an empirical demonstration wherein aggregate approach fared as well as the components approach, but failed to offer any analytical explanation of this paradoxical phenomenon;

4. In India a great debate centring on the money multiplier was seen in the mid dur­ing 1976-78; supporters ofmoney multiplier approach like Gupta (1976a, 1976b) and Swamy (1978) challenged the traditional RBI viewpoint on this approach that the latter had little bearing on the operational aspects of monetary policy; the RBI viewpoint was well articulated in Mujumdar 1976, Shetty, Avadhani and Menon (1976); RBI finally had to accept the multiplier approach in January 1978; there was resurgence of interest in the money multiplier approach on part of RBI as 5.

evidenced by two RBI works Singh, Shetty and Venkatachalam (1982) and Rangarajan and Singh (1984), and an independent work by Chitre (1986); notwithstanding, there were two formidable limitations: (a) the lag pattern in the impact of reserve money on money stock did not receive the attention it deserved, which was particularly surprising because an accurate assessment of the lag structure was the sine qua non of successful monetary management, failure to estimate the money multiplier with reasonable limits of statistical precision might have undesir­able destabilizing influences on money mar­ket conditions, most of the studies discussed above simply ignored the money multiplier lag, and even in those studies like Rangaraj an and Singh (1984), which introduced lags the choice of the lag length tended to be arbitrary, (b) the implicit assumption of uni­directional causality in the sense of Granger (1969) from reserve money to money stock was not always true, but, empirical studies

often indicated possibilities of feedback from money stock to reserve money like Chitre (1986) in the Indian context; the presence of feedback was caused by the presence of the common component currency both in money stock and reserve money and by some other variable(s) affecting both of reserve money and money stock; the former effect would manifest itself as money stock (Granger) c ausing reserve money, the latter effect oper­ated more subtly and could be inferred if the innovations in the two time-series displayed a contemporaneously correlated structure; in presence of such feedback a model could probably generate misleading conclusions, e.g. the dependence of money multiplier on the monetary base in Singh, Shetty and Venkatachalam (1982) was suspected to be an outcome of the failure to model this feedback;

Empirical study on money-income and money-price causality in India, was still in its infancy till the nineties.

extent some by Sims (1972); researchers in the 1970s began developing two-variable causality models; in the SAARC countries, however, the use of bivariate causality models could be traced back to mid 1980s; as an alternative to traditional econometric system of equations in which variables were arbitrarily labelled as endogenous or exogenous, Vector auto-regression (VAR) models emerged as powerful multivariate models since the early 1980s. Above are the studies in the pr-crisis period.

Regarding the econometric tools applied in the above works, it was found that Jha and Donde (2002) and Ahmed (2003) employed vector auto regression (VAR) models accompanied by error correction mechanism (ECM) and Johansen- Juselius procedure; others like Rangaraj an and Arif (1990) employed simulation models containing regressions equations of variety of forms simple linear function and double logarithmic function, and also autoregressive equations of first order (AR₁) and only Ray and Namboodiri (1988) employed filters for pre-whitening purpose i.e. making a non-stationary series stationary. The filter technique did not seem to be popular. Even Jha and Donde (2002) employed ADF test in order to detect the level of integration of the series and accordingly took measures to ensure stationarity.

8.