This Selected Issues paper on Korea provides background information on economic developments and policies, with particular emphasis on 1995–96. Following two years of rapid expansion, led by buoyant investment and exports, economic growth moderated in late 1995 and the first half of 1996. The moderation was in response to the earlier tightening of monetary conditions and less favorable short-term export prospects. The slowdown was reflected in a sharp deceleration in final domestic demand, whose contribution to growth fell from 9.1 percent in 1995 to 6.6 percent in the first half of 1996.


This Selected Issues paper on Korea provides background information on economic developments and policies, with particular emphasis on 1995–96. Following two years of rapid expansion, led by buoyant investment and exports, economic growth moderated in late 1995 and the first half of 1996. The moderation was in response to the earlier tightening of monetary conditions and less favorable short-term export prospects. The slowdown was reflected in a sharp deceleration in final domestic demand, whose contribution to growth fell from 9.1 percent in 1995 to 6.6 percent in the first half of 1996.

II. The Stability of the Demand for Money in Korea 1/

1. Introduction

Since 1979, monetary policy in Korea has been formulated on the basis of an annual objective for the growth of broad money (M2). This approach has worked effectively, and an impressive track record has been established of target ranges for M2 growth being achieved and of inflation declining in line with the medium-term objective of lowering it to rates prevailing in partner countries. A key element in the success of this approach appears to have been a reasonably close relationship between money and nominal income.

There is concern, however, that with institutional changes in the economy, especially in the financial sector, historical relationships among economic variables may be changing. A program of financial-sector reform has been gradually implemented since the early 1980s, and plans for the intensification of reforms were announced in 1993. The program is an ongoing effort, with the full effects of some of the already-implemented measures still to take hold and with further measures on the agenda. The possibility that traditional relationships between the pattern of money holdings and trends in income, prices, and interest rates are being altered has been acknowledged in the setting of monetary policy. In recent years the target ranges for money growth have been quite wide; overshooting of the ranges has sometimes been permitted during the year; 2/ and, while primary attention continues to be paid to M2, a range of other indicators--including interest rates and other monetary aggregates--have begun to be monitored more closely.

The study presented in this chapter empirically examines the relationship between the demand for money and its traditional determinants during the period 1980-95. Among monetary aggregates, the chapter’s main focus is on M2, given its important role in the policy process, but other aggregates are also discussed. The chapter concludes that although there is a reasonably close relationship between M2 demand and its determinants, key parameters may have begun to change in recent years. The evidence is not definitive but--given that the intensification of financial reforms is quite recent and that further reforms are anticipated--it lends support to the view that it will become increasingly important when judging monetary developments to supplement the emphasis on M2 with attention to a range of monetary and financial indicators.

The chapter is organized as follows. Section 2 provides background for the subsequent discussion. After briefly describing some general conclusions reached in the literature about the effects of financial liberalization on demand for money, it presents a brief overview of financial reforms in Korea since the early 1980s and of developments in the velocity of money. Section 3 describes the results of an empirical examination of stability in the demand for money in Korea, in terms of tests for cointegration and parameter stability. Section 4 summarizes the main conclusions of the chapter.

2. Financial liberalization and money demand

a. Background

Economic reforms, especially financial-sector reform, often lead to changes in the relationship between money demand, income, and interest rates. 1/ The changes in the money-demand function can be manifested in several ways. Financial liberalization may lead to sudden shifts in money demand; the values of key parameters (for example, income and interest elasticities) could change substantially or become more volatile; and particular variables may gain or lose statistical significance over time.

The experience of a number of countries indicates that key financial reform measures, such as interest-rate deregulation and measures to improve the functioning and increase the depth of financial markets, may often lead to instability in money demand. 2/ For example, liberalization of interest rates on saving deposits, followed by a rise in these rates, could lead to an increase in the demand for broad money at the expense of narrow money at a given level of income. Measures to improve the functioning and increase the depth of financial markets could lead to both portfolio shifts and changes in the income and interest elasticities of money demand. For example, measures to promote competition among financial institutions often lower transactions costs, causing money demand to respond more rapidly to changes in interest rates. More generally, measures to promote financial market development could result in the availability of new financial assets, which may lead to gradual portfolio shifts away from bank deposits and into the nonbank sector or other assets such as stocks and bonds. On the other hand, measures to strengthen the financial sector and make it more efficient, and to integrate informal credit markets with the formal financial sector, could lead to portfolio shifts from informal markets into the formal financial sector, including bank and nonbank institutions.

b. Developments in Korea

i. Domestic financial-sector reform

During the 1960s and 1970s, the financial system in Korea was an instrument of industrial policy, with government regulations and incentives designed to direct credit to preferred sectors. In addition, interest rates were administered and the government directly intervened in bank management. All of these controls, in combination with the prevalence of policy lending, added up to a tightly controlled financial system characterized by chronic excess demand for credit.

A program of gradual financial-sector reform was begun in the early 1980s, and plans for an intensification of the program were announced in 1993. 1/ The main objective of the reforms has been to increase the efficiency of the financial sector by reducing government intervention. Key measures have included the deregulation of interest rates, an easing in government control over financial institutions, and moves toward indirect instruments of monetary policy (Table II-1). 2/ While the reforms have been wide-ranging in intent, the effects of several measures have not yet been fully reflected in the relevant economic variables. 3/

Table II-1.

Korea: Domestic Financial Sector Liberalization--Schedule and Implementation of Selected Measures 1/

article image
Source: Information provided by the Korean authorities.

The measures are presented according to the original implementation schedule; dates in parentheses indicate month and year of implementation, which, in some cases, was ahead of schedule.

A brief summary of reforms thus far is as follows. A series of steps toward interest rate deregulation during the 1980s freed interest rates on a number of money-market instruments. A more comprehensive liberalization of lending and deposit rates in 1988, however, was halted soon after it began and reversed. In subsequent years, efforts in this area were resumed, and at present most interest rates are formally deregulated. In practice, however, while some rates--such as in the corporate-bond market--tend to fluctuate, bank lending and deposit rates have remained sticky after liberalization. The stickiness may indicate continued informal “window guidance” to influence the setting of bank rates, or may reflect other, unexplained, structural factors impeding flexibility; the reasons are not clearly understood.

The degree and scope of government intervention in credit allocation by financial institutions were scaled back, especially as concerns banks, which were the most restricted among financial institutions. In the 1980s, commercial banks were privatized, several new banks were established, and restrictions on bank and nonbank financial institutions (NBFIs) eased. While both NBFIs and banks have been subject to government controls, NBFIs have traditionally had greater autonomy in their operations and competition among them has been greater than among banks. This has led to rapid development of the nonbank financial sector, evidenced by an increase in its share in total deposits from one-fifth in the mid-1970s to three-fourths by 1995.

Measures were also instituted to check irregular financial practices. Important among them was the introduction in 1993 of the “real-name financial system.” This disallowed the then-prevalent practice of undertaking financial transactions under assumed names. Due to the nature of the transactions it was intended to encompass, however, the extent to which this measure has taken hold is not very clear.

Finally, the authorities have increased their reliance on market-based instruments of monetary control. Recent reforms have developed further the indirect instruments typically found in other countries: open-market operations, reserve requirements, and use of the discount window of the central bank. However, direct instruments--such as mandatory placements of “monetary stabilization bonds” by the central bank--are still used.

Throughout the period, an informal credit market--the “curb market”--has existed, although its importance has declined over time. 1/ In the 1960s and 1970s, with tightly-controlled interest rates and much lending from financial institutions government-directed, there was an excess demand for credit, which manifested itself in a thriving curb market. Reflecting the excess demand, there was a substantial premium on interest rates in the curb market compared with bank deposits (Chart II-1). Since the early 1980s, and especially since the brief interest rate liberalization of 1988, the curb-market rate and the interest premium have declined sharply. 2/ The reduction of activity in the curb market may have been the result of financial sector reforms, which led to some crowding out of the curb market in favor of greater activity in the formal financial sector. 3/ That the curb market still exists, however, suggests that credit rationing may still play some role at the margin.



Citation: IMF Staff Country Reports 1996, 136; 10.5089/9781451822021.002.A002

Sources: Data provided by the Korean authorities.
ii. Trends in velocity

An increase in deposits in the formal financial sector was reflected in a strong growth in the share of monetary aggregates in GDP (see box); that is, in a trend decline in the income-velocity of money (Chart II-2). 4/ The decline since the early 1980s was most marked for the velocity of M3, which fell by two thirds. In 1994-95, the velocities of M2 and M3 remained roughly unchanged, while that of Ml increased. Notably, these developments were not in line with the trend decline observed during the historical period.



Citation: IMF Staff Country Reports 1996, 136; 10.5089/9781451822021.002.A002

Sources: IMF, International Financial Statistics and Bank of Korea, Monthly Statistical Bulletin.1/ Calculated as the ratio between real GDP valued at consumer prices and the relevant monetary aggregate.

Korea: Composition and Developments of Monetary Aggregates

The main monetary aggregates in Korea are M1, M2, and M3. In addition, in 1996 increasing attention began to be paid to the aggregate MCT.

M1 includes currency in circulation and demand deposits held at deposit money banks.

M2 consists of M1 plus time deposits, savings deposits, and residents’ foreign currency deposits held at deposit money banks. 1/

MCT comprises M2 plus certificates of deposits and trust accounts.

M3 is the broadest aggregate. In addition to M2 it includes trust accounts and other deposits of nonbank financial institutions, as well as debentures issued, commercial bills sold, certificates of deposit, repurchase agreements, and, since July 1994, cover bills.

During the past two decades, all of the above monetary aggregates grew in relation to GDP, with the increase being most marked for M3. The share of M3 rose from under 40 percent in the mid-1970s to over 150 percent in 1995; that of M2 rose from under 30 percent to 40 percent; and that of M1, while increasing slightly, remained at about 10 percent. Partial liberalization of the financial sector since the early 1980s contributed to a strong expansion of less regulated financial instruments that are excluded from the M1 and M2 aggregates.

In 1996, starting in May, economic agents began shifting out of trust accounts and into bank deposits, following an increase in the statutory minimum maturity of trust accounts. While the portfolio shifts led to a pickup in the growth rate of M2, they did not affect MCT and M3, which internalized the shifts.

1/ A narrower, more transaction-oriented version of M2 is called M2A. It excludes long-term time deposits and savings deposits with maturities of more than two years. Another less commonly referred to aggregate is M2B. M2B consists of M2A plus certificates of deposit and repurchase agreements of deposit money banks, as well as short-term liabilities of NBFIs.

The trend decline in the velocities of the broader monetary aggregates reflected several factors. First, given that at the outset the financial sector was not as advanced as the rest of the economy, the trend decline may well have been due to monetization of the economy as the financial sector developed. That is, it may have reflected a pattern common among countries with less-developed financial sectors which undertake financial reforms, where money growth often exceeds income growth--there is “financial deepening”--as deposits in the financial system grow. As financial development advances, the downward trend in velocity is typically reversed with financial market innovations permitting agents to economize on their money holdings. 1/ Second, (as mentioned above) use of the curb market appears to have declined, with financial intermediation increasingly taking place through the formal financial sector. Third, the overall rate of national (and private) saving increased substantially during the past two decades, from 22 percent (and 18 percent) of GDP in the 1970s to 35 percent (and 27 percent) in the first half of the 1990s. The rise in saving was reflected in a significant increase in financial assets, including deposits.

3. Estimates of money-demand functions

The key concern motivating the empirical investigation of money demand was to determine whether the demand for the main monetary aggregates--especially M2, in light of its importance in the current policy framework--is a “stable” function of the conventional determinants (incomes and interest rates). 2/ Ideally, a stable money-demand function would involve a well-defined long-term relationship between money and its determinants as well as stability of its parameters across subperiods. Accordingly, the estimation procedure in the present study consisted of two parts. First, the existence of a long-run relationship between money and its determinants was investigated for each of the main monetary aggregates. Then, focussing on M2, key parameters of the money demand function were examined across sub-periods. Given the intensification of the financial reform effort in 1993, it seemed natural to divide the sample accordingly.

a. Specification

It is now common when estimating money-demand functions to employ the methodology of cointegration. 1/ Cointegration is attractive for two main reasons. First, many time series used in economic analysis, including in the analysis of money demand, are unit-root nonstationary. Regression of one unit-root nonstationary series on another can give spurious results and lead to incorrect statistical inferences. 2/ While many series can be rendered stationary by taking their first differences, this would mean giving up information about long-run economic levels of variables. If, however, the nonstationary series are cointegrated--that is, one or more linear combinations of them are stationary--they can, in fact, be included in a regression together, and the long-run information they contain is thereby retained. The second, and here more important, reason that the cointegration methodology is attractive is that a cointegrating relationship among variables can be interpreted as the long-run equilibrium relationship among them. 3/ If, for example, real money balances, real income, and the opportunity cost of money have a cointegrating relationship, then even if they experience a shock that drives them out of equilibrium they will eventually return to this long-run relationship.

Given the relevance of this interpretation for the present study, the first step was to determine whether a cointegrating relationship could be found for the demand for money. While our main interest was in M2, the demand for some other monetary aggregates was also examined.

In the specification of the demand function for money, the study began by following the traditional practice of describing real money balances as a function of a scale and an opportunity-cost variable. The scale variable was represented by real GDP, while two different specifications were considered with regard to the opportunity-cost variable. In the first specification, expected inflation was used as the opportunity-cost variable, to represent the value of the goods that are forgone when money is held. 4/ In the second specification, an interest-rate variable was used; specifically, the difference between the interest rate on alternative assets and on the monetary aggregate in question was considered to represent the opportunity cost of holding money. In addition to the above variables, the specified money-demand function included a time trend. Several recent papers have argued that the inclusion of a time trend can be helpful in capturing the effects of financial innovation, especially when such innovation is characterized by gradual but continuous improvements over time in the techniques for managing money holdings. 1/ In the present study, the trend term was also intended to capture the ongoing, gradual decline of the informal financial sector in favor of formal bank and nonbank financial institutions. 2/

The money-demand equations were specified as:


In the above equations, (M/P) and (Y/P) represent the logarithms of real money balances and real GDP, respectively; πe represents expected inflation (measured as the first difference of the logarithm of the price level), i the interest-rate variable (details of its construction are reported in Appendix 1), T the time trend, and ϵ the error term. 3/

As a preliminary exercise--in order for it to be legitimate to include the above series together in a regression--it was necessary to test that the series were similarly integrated. Based on unit-root tests, it was concluded that all variables were integrated of order 1: that is, their levels were nonstationary, but their first differences were stationary. (Further details of these tests are reported in the Appendix to this chapter.)

b. Estimation results

i. Cointegration

Equations (1) and (2) were tested for cointegration during the period 1980-95, using Johansen’s cointegration technique. For each of the monetary aggregates, a cointegrating relationship could not be rejected (Table II-2). In most cases the results indicated that the cointegrating equations were unique. While some of the results were tentative--for example, the acceptance of cointegration in M2 was sensitive to a correction for the degrees of freedom, and the results in general were sensitive to the inclusion of a time trend and to the lag structure--the broad picture that emerged for each of the monetary aggregates was of a relatively close relationship between money demand and its determinants. 1/

Table II-2.

Johansen Tests for Cointegration, 1980-95

article image
Source: Staff calculations.

The tests were conducted using the levels of monetary aggregates (real terms) expressed in logarithms.

At the 5-percent level of significance. The critical value was 42.44.

With reference to the unadjusted likelihood-ratio statistic.

ii. Stability of the demand for M2

The next step was to examine whether the money-demand function, specifically the demand for M2, has been changing in recent years. The purpose of the tests was not to focus on point estimates of the coefficients, but rather to examine whether and how they might have changed over time. The tests consisted of two separate procedures: (i) the money-demand equations for M2 were run for the periods 1980-93 and 1980-95, and (ii) a Chow test was used to test for a structural break in the M2 equations in 1993.

The regressions indicated that while the presence of cointegration was supported for both periods, some of the main parameters had changed (Table II-3). The coefficient on real GDP rose from 0.2 to 0.4, and went from being insignificant to being significant. 2/ The coefficient on expected inflation remained insignificant; that on the interest rate became significant in the extended sample. 1/ 2/ The coefficient on the time trend was broadly unchanged. 3/

Table II-3.

Korea: The Demand for Broad Money

article image
Source: Staff calculations, based on results of cointegration test.

The table reports estimated coefficients; the reported coefficients on real GDP and expected inflation represent elasticities while those on the interest rate and the time trend represent semi-elasticities. An asterisk (*) indicates statistical significance at the 5-percent level.

An asterisk indicates that the null hypothesis of no structural break in 1993 could be rejected at the 5-percent level.

These results, although not conclusive, are in the direction expected. The higher and significant coefficient on real GDP, and the significant interest coefficient, would be consistent with an economy where agents are increasingly responding to market signals. The lack of significance in the coefficient on inflation is, however, surprising. This could be the result of how the variable was measured--the inflation variable as constructed may not capture the true opportunity cost of holding money--for example, only particular components of the price index, whose effects may be diluted in the aggregate, may be important 4/--or it could, of course, simply indicate that the equation was misspecified. The time trend, which captures gradual but steady changes, would not be expected to change suddenly.

A Chow test for a structural break in the M2 equations in 1993 was not conclusive. 5/ It would in fact have been surprising if there were strong evidence of a break in 1993, given that the intensification of reforms in that year had more to do with clarifying plans for the future than with the immediate implementation of measures or significant changes in economic variables.

c. Further discussion

While the results are not definitive they support the conclusion that the pattern of money demand, while remaining broadly stable, may have started to change in recent years. This conclusion would suggest that financial indicators in addition to M2 would also be useful in gauging monetary developments. It would also suggest that past monetary trends, such as the trend decline in the velocity of money, should not be taken for granted in setting monetary policy for the future. 1/ 2/

This general conclusion finds support elsewhere in the literature on Korea. Two recent papers are of particular interest. The above-mentioned study by the BOK (1995) observed that while the velocity of M2 had recorded a trend decline since the late 1970s, the decline had decelerated since the mid-1980s and that velocity had become more volatile since 1990. It was tentatively argued that the slowing of the velocity decline was not transitory, but rather reflected underlying structural changes related to financial liberalization. A more recent paper by Park and others (1996), which is broader in scope than the present paper, also discussed the stability of money demand in Korea. One finding was that while a cointegrating equation could indeed be found for each of the main monetary aggregates during the period 1973-96, there was a deterioration in the stability of the parameters beginning in the early 1990s.

4. Conclusions

While there is evidence of a relatively close long-run money-demand relationship--for M2 as well as for other monetary aggregates--there are signs that key parameters embodied in that relationship may be beginning to change. These changes seem likely to continue, with several financial-sector reforms still to come and the effects of recently enacted measures still to be fully felt. In formulating monetary policy, while continuing to place key emphasis on M2, it will thus be increasingly important to supplement this emphasis with attention to a range of monetary and financial indicators, in line with the direction that the authorities have begun to take.


This appendix discusses the sources, definitions, and stationarity properties of the data used in the study.

Data on M1, M2, M3, reserve money, the interest rate on time deposits (in deposit-money banks) of maturity greater than one year, the interest rate on corporate bonds, and the consumer-price index were taken from the Monthly Statistical Bulletin published by the Bank of Korea. The study used quarterly averages for the monetary aggregates; these averages were calculated based on the monthly data. Data on curb-market interest rates were provided separately, and were compiled by the Bank of Korea. Data on real GDP were taken from the official national accounts.

The interest-rate-differential variables relating to M2 and M3 were constructed as a difference between the curb-market rate and the “own” rate of return for each aggregate. For M2, the own rate was calculated as the rate on time deposits weighted by the share of quasi-money in M2. For M3, it was calculated as a weighted sum of the rate on time deposits and the rate on corporate bonds. The time-deposit rate was weighted by the share of quasi-money in M2, and the rate on corporate bonds by the proportion by which M3 was larger than M2. The corporate bond rate was used here because interest-rate data on relevant M3 assets--notably certificates of deposit (CD)--were not available over a sufficiently long period, and the corporate bond rate has moved closely with the CD rate in recent years. The curb-market rate was used as the interest rate reflecting the opportunity, cost of holding M1 and reserve money.

Augmented Dickey-Fuller unit root tests were used in order to examine the stationarity properties of the series, in particular to establish the order of integration. The series were integrated of order one (i.e., were I(1)): the levels were nonstationary but the first differences were stationary (Table A1). The real GDP and reserve money series were trend-stationary: the levels turned out to be stationary, but once the trend was removed the series were I(1). The analysis in the paper thus treated all variables as I(1).

Table 1.

Augmented Dickey-Fuller Tests for a Unit Root, 1980-95 1/

(Test Statistics)

article image
Source: Staff calculations.

For any variable, the augmented Dickey-Fuller statistic tests the null hypothesis of a unit root against the alternative hypothesis of stationarity. A finding of statistical significance means that the null hypothesis can be rejected.

All series (except interest rates) are in logarithms.

Unit root not rejected at the 5-percent level of significance, unless otherwise stated.

Unit root rejected at the 1-percent level of significance, unless otherwise stated.

Unit root not rejected at the 1-percent level of significance.

Unit root rejected at the 5-percent level of significance.


  • Arrau, Patricio, Jose de Gregorio, Carmen Reinhart, and Peter Wickham, 1995, “The Demand for Money in Developing Countries: Assessing the role of financial innovation,Journal of Development Economics, Volume 46.

    • Search Google Scholar
    • Export Citation
  • Bank of Korea (Research Department), 1995, “Monetary Management Policies and Procedures in Korea,” (Seoul).

  • Bordo, Michael and Lars Jonung, 1987, The Long-Run Behavior of the Velocity of Money: The International Evidence (New York: Cambridge University Press).

    • Search Google Scholar
    • Export Citation
  • Dekle, Robert and Mahmood Pradhan, 1996, “Financial Liberalization and Money Demand in ASEAN Countries: Implications for Monetary Policy,Paper prepared for the ASEAN Conference on Macroeconomic Issues, held in Jakarta.

    • Search Google Scholar
    • Export Citation
  • Edwards, Sebastian, 1988, “Financial Deregulation and Segmented Capital Markets: The Case of Korea,World Development, Volume 16.

  • Eken, Sena, and others, 1995, “Economic Dislocation and Recovery in Lebanon,” (Chapter V: Estimating Money Demand in Lebanon), IMF Occasional Paper No. 120 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Ericsson, Neil and Sunil Sharma, 1996, “Broad Money Demand and Financial Liberalization in Greece,IMF Working Paper 96/62 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Granger, C. and P. Newbold, 1974, “Spurious Regressions in Econometrics,Journal of Econometrics, Volume 2.

  • Park, Won-Uhm, Jae-Ha Park, Choong-Ryul Lee, and Kong-Cho Choi, 1996, “Monetary and Credit Policy in an Open Economy,” (Seoul: Draft paper, Korea Institute of Finance).

    • Search Google Scholar
    • Export Citation
  • Tseng, Wanda and Robert Corker, 1991, “Financial Liberalization, Money Demand, and Monetary Policy in Asian Countries”, IMF Occasional Paper No. 84 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Tseng, Wanda, and others, 1994, “Economic Reform in China: A New Phase,” (Appendix IV: The Demand for Money in China), IMF Occasional Paper No. 114 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

This chapter was prepared by Mr. Vivek Arora (ext. 36211) who is available to answer questions.


See Chart I-1 in Chapter I.


There is, of course, a large literature on this subject. For a discussion of the key issues and the experience during the 1980s in eight Asian countries, including Korea, see Tseng and Corker (1991).


See Dekle and Pradhan (1996) for a discussion of the recent experience in selected Asian countries.


See SM/94/229 (8/29/94) for a discussion of Korea’s financial-sector reform program and its potential implications for monetary policy formulation.


See SM/95/264 (10/10/95) for an overview of financial-sector reforms in Korea.


The reform process is expected to continue, with several measures remaining on the agenda. See SM/95/209 (8/18/95) for a discussion of the main measures.


Data relating to the curb market are very limited, owing to the nature of the market.


There is, however, still likely to be a large stock of assets in the curb market, reflecting the importance of this market in the past. In 1994, independent estimates suggested it was over 10 percent of GDP.


Edwards (1988) analyzes the extent to which financial deregulation may have led to crowding out of the curb market in the early 1980s.


Velocity was estimated as real GDP multiplied by the consumer-price index divided by the money stock. If the GDP deflator were used instead of the CPI, the overall pattern would not change much except that the velocity of Ml would show a slight upward trend; the analysis in Chapter I, Section 2 employs this definition of velocity.


Reserve money was also included in the analysis, since it is an important aggregate and is the most “controllable” by the monetary authorities. Its inclusion was not with a view to suggesting it as a target, but rather to see whether the stability properties of M2 were shared by some other monetary aggregates.


See Granger and Newbold (1974) for a discussion.


In some cases, the estimation results may indicate more than one cointegrating relationship. While this can potentially complicate matters, since selection of a particular relationship as the representative one is not a straightforward exercise, researchers tend to use theoretical priors as well as statistical criteria to choose among the possible relationships.


Recent examples of studies that use this specification include Eken and others (1995), and Tseng and others (1994).


The sign of the coefficient on the trend term would depend on which of these effects dominated: the improvements in money-management techniques would indicate a negative time trend (as in Arrau et al. (1995)), while the integration of the informal sector into the formal economy would indicate a positive trend.


The above equations impose homogeneity of degree one in prices.


The Johansen test statistic, if not corrected for degrees of freedom, can sometimes lead to cointegration being too easily accepted. Since the Johansen statistic usually reported by most computer programs does not make an appropriate correction, this is done manually as: Adjusted statistic = (raw statistic divided by sample size) times (sample size minus number of variables times number of lags).


The estimate, but not the direction of change, was sensitive to the lag structure that was adopted. While the reported estimates are based on 4 lags, the adoption of, say, only 2 lags would have yielded coefficients on GDP of 1.3 and 1.7 in the two periods.


By way of comparison, a study by the Bank of Korea (1995) covering the period 1983-94 reported coefficients for real income and the interest rate of 0.23 and -0.01, respectively.


The coefficient on the interest rate may appear unduly small. This is because the interest-rate series that was used was in percentage rather than decimal terms (i.e., interest rates were entered as 10 percent, say, rather than as 0.10), which depressed the coefficient correspondingly. It is useful to express the coefficient, whose value was -0.008, in another way: all else equal, a 1 percent rise in the interest rate leads to a 0.8 percent decline in money demand.


Finally, the constant term (not shown in the table) in the two sub-periods was 2.1 and 0.5, respectively, for equation (1) and 2.2 and 1.3, respectively, for equation (2).


See Dekle and Pradhan (1996) for a discussion of this point.


The likelihood-ratio statistic did, however, indicate a break in equation (1). Overall, the results did not lend themselves to a strong interpretation, both because the test statistics were close to the critical values and because the number of observations after the proposed break was small.


An out-of-sample forecast for M2 for the period 1994-95, based on the coefficients estimated for the period 1980-93, showed that actual money demand was lower than forecasted by the model--that is, velocity was higher than forecasted.


In addition, as noted above, velocity developments in 1994-95 were not in line with the trend decline observed during the 1980-93 period.

Korea: Selected Issues
Author: International Monetary Fund