통화정책 자료를 활용한 시장참가자의 기대금리에 관한 연구

김재필,박재진 한국지급결제학회 2023 지급결제학회지 Vol.15 No.1

The contents of press release relating to the monetary policy decision of the central banks have an significant effect on the expectation of the market participants for the future economic condition. In this respect, Moniz and Jong(2014) had studied the effect of the monetary policy communication of the Bank of England on the future market interest rate expectation of the market participants using the economic sentiment analysis. In detail, they had extracted the monetary policy themes, relating to economic growth, inflation, and market interests, from the minutes of the monthly monetary policy meeting of the Bank of England, and they had also extracted the economic sentiment words, such as positive or negative words among those monetary policy themes. Then they had created the economic sentiment index using these words. And with these data, they predicted the market interest rate, such as forward interest rates, using the various prediction models. Finally, they showed that the market interest rate prediction using these data reflected the significant change in the future interest rates. This paper also tried to reveal the usefulness of the sentiment analysis for forecasting of the market interest rates, supposed that the market participants would predict the future market interest rates using those information. Therefore, based on the Moniz and Jong(2014)’s technique, we studied if Korean market interest rates in the future would properly reflect those economic sentiments. For this purpose, this study, firstly extracted five monetary policy themes from the press releases of the Bank of Korea relating to the monetary policy decision in the every month during the period of 15 years, using the text mining method of Park and Kim(2022). And secondly, we extracted economic sentiment words from each monetary policy themes and created each economic sentiment index. Lastly we forecasted the future interest rates of a Convertible Deposit(maturity of three months) and Treasury Bond rate(maturity of three years) in one month later just after the date of the monetary policy decision of the Bank of Korea. As a forecasting method, we used machine learning method, such as support vector machine and random forest, and a deep learning method using simple neural nets in addition to a general linear regression method. And we included the average of the above all predicted interest rates as a predicted data. For the test of forecasting power, we segregated the whole data into two parts. One part is 75% of the whole data for the training of machine learning and deep learning models, and the other is 25% of the whole data for the test of the forecasting power. As a result of this analysis, we found that the interest rate forecasting power were much higher when the economic sentiment information was included. Among the forecasting methods, the machine learning methods had much higher forecasting power. Especially the random forest method had the better forecasting power among all forecasting methods. Lastly, the average of all forecasting results of every forecasting methods had the best forecasting power. As a result of this study, it is implied that the usage of economic sentiment analysis would be a better way of forecasting the future interest rates and the machine learning methods would be also helpful to forecast future interest rates. In addition, it is also implied that the sentiment analysis may be helpful for the study of forecasting the other market variables such as exchange rates and stock prices. 중앙은행의 통화정책 결정관련 내용이 향후 시장참가자의 시장금리에 대한 기대에 매우 큰 영향을 미친다는 점에서 Moniz and Jong(2014)은 경제적 감성 분석기법을 이용하여 중앙은행의 통화정책관련 커뮤니케이션이 시장참가자의 기대금리에 미치는 효과를 연구하였다. 구체적으로 매월 영란은행의 통화정책위원회 회의록(Monetary Policy Committee Minutes)에서 통화정책 주제(경제성장, 인플레, 시장금리)를 추출하고, 동 주제에서 경제적 감성(economic sentiment) 단어를 선별하여 경제적 감성지수를 산출한 다음, 이들을 합성한 시계열로 시장참가자의 미래 시장금리 기대 수준을 예측하였다. 다양한 모형에 의한 금리 예측과 함께 Ensemble 기법에 의한 예측치를 기대금리 수준으로 평가한 결과 예측력이 높아짐을 보여주었다. 본 연구는 Moniz and Jong(2014)의 연구 방법을 원용하여 우리나라의 통화정책 관련자료를 사용하여 시장참가자의 향후 기대금리 수준에 대한 예측을 실시하였다. 먼저 우리나라 통화정책 커뮤니케이션 효과분석에 관한 기존 연구(박재진‧김재필, 2022)의 방법을 이용하여 통화정책관련 기자간담회 자료에서 통화정책 주제를 추출한 후, 이들 각 주제에서 경제적 감성단어를 선별하였다. 그리고 경제적 감성단어를 반영한 새로운 통화정책 주제 시계열을 작성하여 향후 시장 금리수준을 예측하였다. 예측모형으로는 일반선형회귀분석모형 이외에 random forest 등 기계학습(machine learning) 모형과 인공신경망(deep neural net) 모형도 사용하였다. 예측대상금리는 매월 통화정책결정 이후 1개월 후의 시장금리(CD금리, 3년만기 국고채금리)로 하였으며, 이를 위해 총 15년간의 시장금리 시계열 자료 중 기계학습 모형의 학습을 위해 동 시계열의 75%를 사용하고 예측력 검증을 위해 나머지 25%를 활용하였다. 감성분석 효과를 살펴보기 위하여 감성분석기법을 사용한 경우와 사용하지 않은 경우로 구분하여 예측력을 비교하였다. 분석 결과 감성분석기법을 사용한 경우의 예측력이 훨씬 높아지는 것으로 나타난 가운데 예측 모형 중에서는 기계학습 모형에 의한 예측치가 상대적으로 높은 예측력을 보이는 것으로 나타났다. 이러한 실증분석 결과는 우리나라 통화정책과 관련하여 시장참가자가 기대하는 향후 시장금리 수준을 예측하고자 하는 경우 통화정책 주제 외에 경제적 감성단어에도 주의를 기울일 필요가 있음을 시사한다. 또한 금리 예측 시 일반선형회귀모형 외에 기계학습 모형도 적극 활용하는 것도 바람직함을 보여준다. 이와 같은 연구결과에 비추어 향후에는 금리 외에도 환율, 주가 등 시장변수 예측에 관한 연구에 있어서 경제적 감성분석을 활용한 연구가 활성화될 것으로 기대된다.

미디어 뉴스의 통화정책 전달효과분석

김재필,박재진 한국지급결제학회 2022 지급결제학회지 Vol.14 No.2

Jae J. Park and Jaepil Kim(2022) had studied the short term effect of Korean monetary policy communication effects of Korea using the text mining techniques with the press releases relating to the monetary policy decision on the day of the monetary policy meeting of the Bank of Korea. This is the follow-up study of the previous study, using new information such as the media news texts named as Social Metrics which collects all the media news texts in Korea and provides with a bag of the words relating to a subject or a certain word concerned. This paper used the same analytic techniques of the previous study to find out the monetary policy transmission effects of media news relating to the monetary policy of Korea on the days of the monetary policy meeting of the Bank of Korea. The result of this study revealed the distinct difference between the previous study and this study. Firstly, the number of words relating to the monetary policy themes extracted from the media news texts was 3,369. It is l.8 times larger than that of words(1,886) in the previous study. Secondly, the correlation analysis between the series of the monetary policy themes with the current and future economic conditions discovered that those series extracted from the media news texts had much stronger correlations than those in the previous study. In detail, the number of statistically significant variables among the variables of the monetary policy themes correlated with those variables of current and future economic conditions in this study were 32 which is much larger than the number of 19 in the previous study. Lastly, the regression analysis of the independent variables of the series of the monetary policy themes and the dependent variables of the daily change in five market interest rates (Convertible Deposit, Commercial Paper, Monetary Stabilized Bond, Treasury Bond, Corporate Bond) in this study showed that the number of statistically significant coefficients of those independent variables in this study were numbered 1 to 4 larger than those in the previous study respectively except Convertible Deposit rate. These results suggests that the media news texts should have much larger information on the Korean monetary policy and economic conditions than the press releases of the Bank of Korea used in the previous study. As an implication of this study, the media news texts could be used to analyse the short term effects of the Korean monetary policy in addition to the previous study. And it also implies that new texts such as SNS(social network services) data, having the view of various economic agents on the monetary policy and the Korean economic conditions, could be used for the additional study on the transmission effects of monetary policy of Korea. 본 연구는 국내 언론사의 통화정책관련 뉴스가 경제 여건과 시장금리에 어떤 영향을 미치는지를 분석하였다. 이를 위해 언론사 뉴스를 문서로 재구성한 Social Metrics 자료(이하 SM 자료라 함)를 이용하여 선행 연구(박재진‧김재필: 2022)와 동일한 기법으로 실증분석을 하였다. SM 자료로 추출한 통화정책 주제와 주요 경제변수와의 상관관계를 분석하고, 동 주제를 독립변수로 하여 시장금리 변동과의 회귀분석을 실시하여 그 결과를 선행 연구와 비교하였다. 먼저 상관관계 분석결과를 보면 SM 자료를 사용한 경우 통화정책 주제가 총 32개의 경제변수(경제동향변수 12개 및 경제전망변수 20개)와 상관관계가 있는 것으로 나타나 기자간담회 자료를 사용하는 경우(총 19개의 변수: 경제동향변수 8개와 경제전망변수 11개)보다 훨씬 많은 것으로 나타났다. 또한 5종류의 시장금리(CD, CP, 통화안정증권, 국고채, 회사채) 변동에 대한 통화정책 주제와의 회귀분석결과에서도 SM 자료를 사용하는 경우가 기자간담회 자료를 사용하는 경우보다 통계적으로 유의한 변수의 수가 각 금리별로 1∼4개 많은 것으로 나타났다. 구체적으로는 단기금리인 CD금리의 경우 두 자료에 의한 주제에 대해 비슷한 반응을 보였으나 나머지 금리는 기자간담회 자료보다는 SM 자료에서 추출한 통화정책관련 주제에 대해 더 뚜렷하게 반응하는 것으로 분석되었다. 이처럼 기자간담회 자료보다 SM 자료에 의한 실증분석 결과가 훨씬 더 뚜렷한 결과를 보인 것은 SM 자료에서 추출한 단어수(3,369개)가 기자간담회 자료에서 추출한 단어수(1,886개)보다 약 1.8배의 크기를 가짐에 따라 통화정책과 경제 여건에 관한 정보를 훨씬 더 많이 포함하고 있기 때문인 것으로 풀이된다. 따라서 우리나라 통화정책의 커뮤니케이션 효과 분석을 위해 기존 연구결과와 함께 SM 자료에 의한 연구결과를 추가로 사용함이 바람직할 것으로 보인다. 또한 향후 동 효과를 보다 정밀하게 분석하기 위해 다양한 Social Network Service 자료를 활용한 연구가 활성화될 필요가 있다고 판단된다.

Pre-processing of Photoplethysmographic Waveform for Amplitude Regularization

김재필,이정환,신항식 대한전기학회 2019 Journal of Electrical Engineering & Technology Vol.14 No.4

This study is aimed to develop a post-processing method to regularize amplitude of PPG waveforms and evaluate the performance of the algorithm. The proposed procedure for regularizing amplitude consists of extracting the signal envelopes based on time-domain fi ltering, generating a compensation curve, and reconstructing the waveform. The proposed technique was evaluated with respect to dispersion and waveform fl uctuation by means of coeffi cients of variation (CV) and a scaling exponent of detrended fl uctuation analysis (DFA). Other criteria included the detection performances of major features such as upper and lower fi ducial points and maximum dV / dt . In developing this method, we used a clinical dataset that comprised 74,225 beats. By applying reconstruction algorithm, the %CV and the scaling exponent of DFA were signifi cantly decreased from 115.4 to 71.4 and from 1.029 to 0.836, respectively in peak amplitude. In valley amplitude, they were also decreased from 38.3 to 13.2 and from 0.993 to 0.713, respectively. Moreover, sensitivity of maximum dV / dt point, upper and lower peak detection are increased by 8.50, 2.72 and 5.21% after reconstruction, respectively. Our results clearly showed that the proposed reconstruction technique could reduce amplitude fl uctuation and improve detection of fi ducial points.

Synthesis and Optical Properties of D-π-D Structured red emission DPP dyes containing Triphenylamine donors

김재필,황태규,김정윤 한국공업화학회 2016 한국공업화학회 연구논문 초록집 Vol.2016 No.1

Diketopyrrolopyrrole(DPP)dyes are promising organic colorants with characteristics of intense color, excellent stabilities. Especially, red emitting DPPs have been adopted for numerous industrial applications such as photoconversion film and OLED. High quantum yield(Q.E), large stoke shift and high stabilities are vital properties in such industrial applications. In this study, three kinds of Donor-π-bridge-Donor (D-π-D) structured DPP red emitting dyes were synthesized. Synthesized dyes contained electron enrichment triphenylamine(TPA) moieties with/without adittional electron donating donors. Since the triphenylamines can induce steric hindrance to the entire dye structure, their aggregation can be disturbed. Therefore, higher quantum yield and bathochromic shift can be achieved. The three synthesized dyes containing TPA moieties showed bathochromic shift and large stoke shifts compared to the dyes without TPA moieties. Although, their thermal stability and Q.E were not improved.

u-금융서비스의 비즈니스 모델 및 발전 과제

김재필,이충열 한국정보사회학회 2008 정보사회와 미디어 Vol. No.

<P>본 연구는 u-금융서비스의 현황과 전망을 살펴보고, u-금융의 경제학적 특성을 제시한 후, 이를 비즈니스 모델별로 구분하여 분석하고, u-금융서비스 발전방안을 제시하였다. 분석 결과, u-금융모델은 이동통신사 중심 모델, 금융기관 중심 모델, 개인 대 개인 모델, 협력 모델로 구분되며, 현재 국내외에서 활용되고 있는 u-금융서비스는 대체로 은행 중심 모델로 나타났으나, 향후에는 USIM 및 UICC 등 관련기술의 발전으로 금융기관, 이동통신사, 모바일 기기 제조사 등 참여자들 간의 협력이 필요한 협력모델 중심으로 변화할 것으로 예상되었다. 이에 따라 기술적 안전성확보 및 호환성 개선, 금융기관과 이동통신사 및 기타 IT기업 간의 협력 강화, 인프라 확보 및 공공서비스, 정부기관 간의 협력 강화 등이 향후 해결되어야 할 과제로 제시되었다.</P>

프로펠러의 피치각도와 엔진 RPM 변화에 따른 항공기소음 및 추력 평가

김재필,류종현,김병삼 한국기계기술학회 2022 한국기계기술학회지 Vol.24 No.6

Most domestic pilots are trained at local airfields using propeller aircraft. Training aircraft are mainly trained in the airspace around the aerodrome, and mainly take-off and landing exercises that require a lot of practice among flight control skills. Aircraft noise is a sound that humans do not want. In this study, based on the Rotax 914 engine used in Korea, the propeller blade angle was changed by 1 degree for the 3-leaf “K company” propeller and the 3-leaf GSC wooden propeller, and the engine RPM was changed to examine the noise and thrust changes. The purpose of this study is to check whether noise and thrust loss are the least at the engine's maximum RPM, and to propose an aircraft operation plan in the noisy aerodrome area based on the values.

항공기 엔진의 이상 진동 경고 시스템

김재필,류종현 한국기계기술학회 2021 한국기계기술학회지 Vol.23 No.6

This study indirectly measures the vibration value corresponding to the abnormal vibration generated by the small engine for light aircraft to which the dual carburetor is applied by the pressure difference from each carburetor. It relates to a system for outputting a warning to a pilot, comprising two pressure gauges for measuring the pressure from each carburetor and a warning signal output unit for outputting a warning signal corresponding to the pressure difference measured by the two pressure gauges do.

전자화폐 도입의 경제적 효과

김재필 한국정보기술응용학회 2002 한국정보기술응용학회학술대회 Vol.2002 No.1

염기성 염욕내에서의 캐리어의 작용에 관한 고찰

김재필,S. M. Burkinshaw 한국섬유공학회 1991 한국섬유공학회지 Vol.28 No.8

The effectiveness of carrier as dyeing accelerator in the cationic dye-acrylic fibre system and the role of carrier film formed at the fibre surface were investigated in relation to the solubility of the dye in the carrier. Partition coefficients of three cationic dyes between carrier phase and carrier saturated aqueous solution were determined and taken as the measure of solubility of the dye in the carrier. The effect of hydrophobic carrier film on the rate of dyeing was studied using o-phenol phenol and compared to that of benzyl alcohol as a hydrophilir.

풍경 초상 외 2편

김재필 고요아침 2016 열린시학 Vol.21 No.2