벡터오차수정모형을 활용한 환율과 외환보유액 간 장단기 관계 분석

Recently, it is observed in Korea that the international reservescontinue to grow with the sustained surplus of current account. Major tradingpartners and international organizations raise a suspicion that Korean foreignexchange authorities have intervened in the market from mercantile motives. In order to verify the suspicion, this paper analyzes the VECM(vector errorcorrection model) which shows both short-run dynamics and long-runequilibrium relation between exchange rate and international reserves. Unit-root tests (ADF and PP) are performed for each of ten variables:exchange rate, reserves, inflation, stock price, interest rate, industrialproduction, exports, imports, inbound foreign investments and outboundforeign investments. The results indicate that all individual variables areintegrated of order 1. In addition, the results of cointegration tests (tracestatistic and maximum eigenvalue statistic) indicate that there existcointegrating relations between exchange rate and reserves, and among allten variables as well. Based on those relations, it is allowed to employ VECMsin order to analyze the pattern, effect and motive of intervention. Beside addressing conventional issues such as patters and effects ofintervention in the dimension of short-run dynamics; the model explicitlyincorporates the dimension of long-run equilibrium relation betweenexchange rate and international reserves in order to infer motives aroundinterventions. Main findings are as follows. First, it seems that Korean foreign exchange authorities have frequentlyturned aside from smoothing motives which aim to resolve deviations ofexchange rate from the long-run target level as well as to mitigate short-termfluctuations. It is confirmed by observed asymmetrical patterns and effectsof intervention. Especially, the decrease of reserves reacting to the increaseof exchange rate is not verified, which is not compatible with smoothing motives. Moreover, persistent changes in reserves along with oscillatingchanges in exchange rate cannot be explained though symmetrical against-the-wind interventions from smoothing motives. Secondly, it is not clear whether interventions have been effective. With the result showing that the short-term effect of intervention isasymmetrical, only the dollar-sale-intervention at the previous monthsignificantly contributes to the increase of exchange rate at the currentmonth. The observation does not fall into line with both direction criterionand smoothing criterion by which interventions can be evaluated as successful. The impulse-response analysis with regards to long-term effect, which tracesthe effect of a plus one percent shock to reserves on current and futurevalues of exchange rate, shows that exchange rate drops immediately afterthe shock and its decreased margin is gradually reduced thereafter. It impliesthat exchange rate cannot be intentionally depreciated through interventionsfrom mercantile motives. Thirdly, interventions deviated from smoothing motives are compatiblewith precautionary motives rather than mercantile motives. Such a judgementcan be supported by the estimated cointegrating vectors indicating thatreserves in Korea have maintained a positive long-run relation with exchangerate. When exchange rate increases sharply, Korean foreign exchangeauthorities are not likely to execute dollar-sale-interventions in fear oflosing international reserves. Moreover, dollar-purchase-interventionsfrom precautionary motives against liquidity risk are likely to be frequentlyemployed when exchange rate is high. In order to check the robustness of the judgement that continuouslyincreasing international reserves in Korea can be successfully explainedby precautionary motives, VECMs are applied to each of two sub-periods. The first sub-period is from January 1994 until December 2003. The secondsub-period is from January 2004 until December 2013. The estimation resultsshow that intervention patterns compatible with precautionary motives arenoticeable during the first sub-period. In contrast, observed patterns ofintervention during the second sub-period become close more to smoothingmotives. Also, the analysis on the VECM consisting of ten endogenousvariables shows the essentially same results. Furthermore, key findingsof the analysis remain valid even after interest income on reserves is adjustedand even when the effective exchange rate is used.