Evaluating the Financial Market Function in Prewar Japan using a Time-Varying Parameter Model
TTesting Semi-Strong Form Efficiency of the Prewar
Japanese Stock Market
Kenichi Hirayama a and Akihiko Noda b,c ∗ a Tokio Marine Asset Management Co., Ltd., 18th Floor Tekko Building, 1-8-2 Marunouchi, Tokyo 100-0005, Japan b Faculty of Economics, Kyoto Sangyo University, Motoyama, Kamigamo, Kita-ku, Kyoto 603-8555, Japan c Keio Economic Observatory, Keio University, 2-15-45 Mita, Minato-ku, Tokyo 108-8345, Japan
This Version: August 4, 2020
Abstract:
This paper examines Fama’s (1970) semi-strong form efficient market hy-pothesis (EMH) in the prewar Japanese stock market using a new dataset. We partic-ularly focus on the relationship between the prewar Japanese stock market and severalgovernment policy interventions to explore whether the semi-strong form stock marketefficiency evolves over time. To capture the long-run impact of government policy in-terventions against stock markets, we measure the time-varying joint degree of marketefficiency and the time-varying impulse responses based on Ito et al.’s (2014; 2017) gen-eralized least squares-based time-varying vector autoregressive model. The empiricalresults reveal that (1) the joint degree of market efficiency in the prewar Japanese stockmarket fluctuated over time because of external events such as policy changes and wars,(2) the semi-strong form EMH is almost supported in the prewar Japanese stock market,and (3) the markets rapidly reflect the information of the external events through time.Therefore, we conclude that Lo’s (2004) adaptive market hypothesis is supported in theprewar Japanese stock market even if we consider that the public information affects thestock market.
Keywords:
Efficient Market Hypothesis; Adaptive Market Hypothesis; GLS-Based Time-Varying Model Approach; Degree of Market Efficiency; Semi-Strong Market Efficiency.
JEL Classification Numbers:
C22; G12; G14; N20. ∗ Corresponding Author. E-mail: [email protected], Tel. +81-75-705-1510, Fax. +81-75-705-3227. a r X i v : . [ q -f i n . S T ] A ug Introduction
Since Fama (1970), economists have been examining the efficient market hypothesis(EMH) in various stock markets. There are three forms of the EMH: the weak-form,semi-strong form, and strong-form. Most of the previous studies examine the weak-formand the semi-strong form EMH in the stock markets; but almost none examine the strong-form EMH. This is because there is no arbitrage opportunity even if they can utilize theinside information. If the strong-form EMH is true, investors do not have an incentive fortrading. The weak-form EMH is the most well-known hypothesis that asserts that cur-rent stock prices only reflect historical prices. As Lim and Brooks (2011) point out, manyrecent studies on the weak-form EMH conclude that stock markets are almost efficientbut market efficiency changes over time. This means that there is a strong possibility thatan evolutionary alternative to the EMH, the adaptive market hypothesis (AMH) of Lo(2004), is supported in the stock markets. The AMH reinforces the view that the marketevolves over time, as does market efficiency. It implies that market efficiency can arisefrom time to time due to changing market conditions such as behavioral bias, structuralchange, and external events.The semi-strong form EMH asserts that stock prices adjust rapidly to the releaseof the price around new public information. We consider that new public informationincludes not only corporate financial information but also information on governmentpolicy interventions. In the view of corporate financial information, several studies showthat financial ratios and macro variables are useful in predicting stock returns, and theyconclude that the semi-strong form EMH is not supported (e.g., Fama (1981); Campbell(1987); Campbell and Shiller (1988a,b); Fama and French (1988, 1989)). Studies in thesecond category, on the other hand, shed light on the relationship between the govern-ment policy interventions and stock market efficiency (e.g., Davidson and Froyen (1982);Pearce and Roley (1985); Darrat (1988); Hancock (1989); Bernanke (2004); Ehrmannand Fratzscher (2004); Laopodis (2009)). This is because government policy interven-tions might cause the divergence between fundamental value and the prevailing stockprice, and decline the stock market efficiency. However, whether the stock prices reflectthe information on government policy interventions and whether the semi-strong formEMH is supported or not remain controversial (see Fama (1991), Becker et al. (1996),and Malkiel (2003) for details).Furthermore, several recent studies examine how the central bank’s unconventionalmonetary policy affects the stock market. Chuliá et al. (2010) explore the Federal Re-serve’s announcement effect for the federal funds target rate changes on the stock prices,volatilities, and correlations. They find that the stock market responds differently to pos-itive and negative target rate surprises. Ueda (2012) examines how the Bank of Japan(BOJ)’s unconventional monetary policy strategies affect the stock prices. He concludesthat the strategies except for quantitative easing move the stock prices in the expecteddirections. In December 2010, the BOJ introduced the purchasing exchange-traded funds(ETFs) program and has been rapidly expanding the purchasing scale after April 2013. Afew studies examine the effect on the stock markets because the BOJ is the only centralbank purchasing stocks through ETFs. Ueda (2013) investigates the impacts of the ETFspurchasing by the BOJ and shows that the BOJ’s unconventional monetary policies havepositive effects on the TOPIX and the current exchange rates of the U.S. dollar againstJapanese yen. Harada and Okimoto (2019) examine the cumulative treatment effects onthe Nikkei 225 by the BOJ’s EFTs purchasing program and show the effects on the Nikkei125 have already reached around 20% as of October 2017.As mentioned above, economic policies often cause price distortion in stock marketsand affect the stock market efficiency. We can consider that security prices are almostefficient in the long-run by rapidly reflecting new information on economic policies becausemarket institutions and (information and communication) technologies are developedenough in the present day. However, when those were still developing, the prices didnot always reflect new information on economic policies. In practice, Noda (2019) showsthat the degree of market efficiency in the weak sense changes and evolves over timeusing the prewar Japanese stock market data. Therefore, it is reasonable to expect thatwe consider the market efficiency in the semi-strong sense also changes and evolves overtime. Moreover, we can reveal the effect of policy changes on the stock market under theclosed economy because our dataset includes the wartime period. Particularly, we payattention to the transition process from the open economy to the closed economy in theprewar Japanese stock market. We then focus on the relationship between the prewarJapanese stock market and several government policy interventions to explore whetherthe semi-strong form stock market efficiency evolves over time.Most of the previous studies employ the event study analysis to evaluate the semi-strong form EMH in accordance with Ball and Brown (1968) and Fama et al. (1969). The event study analysis is a very simple and convenient method; however, there is aserious problem. Particularly, we calculate a cumulative abnormal return (CAR) and testwhether the CAR is statistically significantly different from zero on the specified eventwindows. We find that the event affects the stock prices when the CAR is statisticallysignificantly different from zero. We can conclude the semi-strong form EMH is notsupported. Note that the result of the event study analysis provides us estimates of onlythe short-run impact, not the long-run impact. Furthermore, the results of the eventstudy analysis are quite unstable to even small changes in research design as shown inMcWilliams and Siegel (1997). Hence, we adopt Ito et al.’s (2014; 2017) generalized leastsquares (GLS) based time-varying vector autoregressive (TV-VAR) model to investigatethe long-run impacts of public information, especially government policy changes, in thestock markets. In the system of the GLS-based TV-VAR, we estimate the joint degree ofmarket efficiency that is obtained from the time-varying impulse responses and evaluatehow public information affects the stock market efficiency in the semi-strong sense.This paper is organized as follows. Section 2 provides a brief literature review thatexplains this paper’s significance in finance and economic history. Section 3 presents ourempirical method for estimating the joint degree of market efficiency as the measurementof the semi-strong market efficiency that is based on the Ito et al.’s (2014; 2017) GLS-based TV-VAR model. Section 4 presents our datasets that include new data of theperformance index for the prewar Japanese stock and government bond markets estimatedby Hirayama (2017a,b, 2018a,b, 2019, 2020a,b) and presents some statistical test results.Section 5 shows our empirical results using a GLS-based TV-VAR model and discussesthe relationship between policy changes, major historical events, and time-varying semi-strong market efficiency in the prewar Japanese stock market. Section 6 concludes thepaper. We call the period of prewar and wartime as “prewar” in this paper. See Brown and Warner (1980) and MacKinlay (1997) for technical details. Literature Review
This study examines Fama’s (1970) semi-strong form EMH in the prewar Japanese stockmarket using Ito et al.’s (2014; 2017) GLS-based TV-VAR model. We expect that ourstudy provides useful insights into the above-discussed important issues of finance andeconomic history. Previous research in the field of modern economic and financial historycan be divided into two types: research on the economic system since the Meiji period, andresearch on the history of specific financial institutions and industries. The former focuseson policy changes and flow of funds. Fujino and Teranishi (2000), Utsunomiya (2013),and Shibata (2011) systematically depict the structure of flow of funds. These studiesalso deal with the flow of funds among economic actors such as governments, households,corporations, and financial institutions. Typical examples of the latter research are Kato(1965a,b) and Kasuya (2006), which clarify financial intermediation systems based onfinancial data of individual financial institutions. These approaches address changes infinancial intermediation mechanisms or structures. However, as Ito (1999) has pointedout, to grasp the actual condition of the financial intermediation system, it is necessarynot only to analyze the structural changes described above, but also to take an approachto reexamine whether the functions of the system are being fully utilized. This is becauseimproving the financial structure does not necessarily mean an efficient financial interme-diation function. For example, in the financial system in developing East Asia in 1990,the financial structure and its function diverged. In this study, we focus on this pointand examine the efficiency of financial markets to clarify whether financial intermedia-tion is functioning. In this study, we focus on the time-varying nature of the semi-strongform market efficiency to quantitatively grasp the functions of the stock market in prewarJapan, which will contribute to complementing conventional studies of economic history.There have been few previous studies on the functioning of stock markets. In previousstudies, economic policy by Minister of Finance Korekiyo Takahashi from 1932 to 1936has been one of the main themes of policy analysis in economic history. Many of thembelieve that the recovery from the global Great Recession became possible by policypackages such as fiscal expansion policies, currency depreciation, and monetary easingpolicies. However, quantitative studies that have analyzed the effects of this policy arelimited to Shibamoto and Shizume (2014). They explore that this policy had a largeeffect on raising expected inflation before seceding from the gold standard system.Studies on financial markets in the early 1930s have focused on the government bondmarket, but not on the stock market. The main monetary policy at that time wasfocused on the government bond market and the foreign exchange market. The interest ofpolicymakers in the stock market was low. In 1932, however, stock prices rose significantlyand the situation in Japan was different from the stagnation of stock indices in Europeand the U.S. Therefore, it is necessary to reexamine the situation quantitatively. Inthis sense, Bassino and Lagoarde-Segot (2015), a financing study on the stock marketbefore and during the war, is a valuable contemporary study. However, the market dataused (daily data in the
Economic Yearbook of the Toyo Keizai ( Oriental Economist ) willneed to be reconsidered. First, the stock price data do not consider that no correctionhas been made to the settlement of rights and additional payments, or the return fromdividends. Second, Japanese government bond (JGB) data are limited to 5% couponbonds and do not consider changes in spreads by interest rate, nor do they include 3.5%coupon bonds issued in large volumes after the outbreak of the Second Sino-Japanese See Appendix for details. Since the restoration of the gold embargo, as can be seen from the examples of theBOJ’s underwriting of government bonds and the revision of the Capital Flight PreventionLaw, Japanese government intervention was strengthened in the government bond andforeign exchange markets earlier than in the stock market. By contrast, in the stockmarket, strong government intervention such as the Share Price Control Ordinance andStock Price Support Organizations was not implemented until the wartime. Quantitativeanalysis of market efficiency will help to identify when government price intervention inthe early Showa period had a significant impact on the market. This is because policyeffects cannot be sufficiently measured only by studying the history of economic policyat that time and by checking trends inflow of funds. Quantitative assessment of marketimpact will also be an effective tool for studying economic history.The three main issues related to market functions in research on economic historyare as follows. First, there is a view that the stock market stagnated because investorsdropped out after the Showa Depression in 1927 and that the stagnation of the marketcontinued even after the World War II (WWII). Another view states that although theprice control of the government bond market was strengthened, the degree of freedom inprice fluctuations in the stock market was secured to some extent. The two views on thestock market could be explored to a certain extent through the time-series evolution ofmarket volatility and comparisons of other assets. Hirayama (2018a) provides descriptivestatistics such as the stock market price volatility in the U.S. and Japan, and comparesthe difference between the markets. Based on the index he calculated, the risk level ofthe stock market in the 1940s (until November 1944) was 11.8% in Japan and 13.5% inthe U.S. Although the risk level of the Japanese stock market is relatively low, it canbe said that a certain degree of volatility has been maintained compared with the U.S.and Japan. Additionally, the yield spread, which is the difference between the yield onthe long-term government bonds and the equity earnings yield, has been rising in bothmarkets, while the relative positions have changed alternately since the late 1930s. Interms of the comparison between Japan and the U.S. in the secondary market, it may notbe said that only the Japanese market has shrunk. In this way, it is possible to supplementthe view of economic history by quantitative bilateral market analysis. Second, whethercorporate finance functioned mainly in the securities market or in the traditional bankingindustry before and after World War I (WWI) remains debatable. Although the views ofthese discussions do not agree, a quantitative understanding of changes in the efficiencyof the stock market and other markets will have certain implications for the discussion.This will contribute to broaden the scope of this discussion by examining the functioningof securities markets.Some studies examine the efficient market hypothesis using the prewar Japanese stockmarket data. Kataoka et al. (2004) investigate whether the prewar Japanese stock mar-ket was efficient using the daily stock prices in 1903 alone. They find that the market isefficient in the weak sense but the market is not efficient in the semi-strong sense. Suzuki(2012) assumes known breakpoints and employs the Choudhry’s (2010) method to ex-amine the semi-strong EMH using the Japanese daily stock market data during WWII.He finds that market efficiency declined after the start of the war and that the knownbreakpoints are consistent with the variation of stock prices. Bassino and Lagoarde-Segot(2015) explore the weak-form EMH in the 1930s’ Japanese stock market using the gener-alized autoregressive conditional heteroskedasticity (GARCH)-in-mean model and argue Additionally, income returns, which account for the majority of total JGB returns, are not consid-ered.
We first introduce Ito et al.’s (2014; 2017) GLS-based TV-VAR model to analyze the time-varying nature of the semi-strong market efficiency on prewar Japanese stock market.We suppose that p t is a price vector of the three securities (stock, government bond, andexchange rate) in t period. Our main focus is reduced to the following condition: E [ x t | I t − ] = , (1)where x t denotes a return vector of the securities in t period. We calculate that the i -thelement of x t is ln p i,t − ln p i,t − for each security. Note that Equation (1) implies that allconditional expected returns of the three securities at t period given the information setavailable in t − period are zero. If x t is stationary, we denote the time-series process of x t using the Wold decomposition as x t = µ + Φ u t + Φ u t − + Φ u t − + · · · = µ + Φ( L ) u t , (2)where Φ( L ) is a matrix lag polynomial of a lag operator, µ is the mean of x t , and u t follows an independent and identically distributed multivariate process with a mean ofzero vector. We assume that coefficient matrices { Φ i } ∞ i =0 are k × k dimensional parameter5atrices. We then compute the impulse-response functions along with the identificationassumptions such as Φ = I . Note that the EMH holds if and only if Φ i = 0 for all i > ,which suggests that the market deviation from the efficient market reflects the impulseresponse, a series of u t . In this study, we construct a relative degree based on the impulseresponse to investigate whether the semi-strong form EMH holds for the prewar Japanesestock market.We can obtain that the impulse response is to employ a VAR model and to alge-braically compute its coefficient estimates. Suppose that the vector return process x t ofthe three securities is invertible under some assumptions. Then, we consider the followingstandard VAR( q ) model: x t = ν + A x t − + A x t − + · · · + A q x t − q + ε t ; t = 1 , , . . . , T, (3)where ν is a vector of intercepts; ε t is a multivariate error term with E [ ε t ] = , E [ ε t ] = σ ε I , and E [ ε t ε t − m ] = for all m (cid:54) = 0 . We can measure a relative degree about the semi-strong form market efficiency that varies over time according to the following procedure.First, we compute a cumulative sum of the coefficient matrices of the impulse response: Φ(1) = ( I − A − A − · · · − A q ) − , (4)We next define a joint degree of market efficiency as the measurement of the semi-strongform market efficiency: ζ = (cid:112) max [(Φ(1) − I ) (cid:48) (Φ(1) − I )] , (5)to measure the deviation from the efficient market. We can understand that in the case ofthe efficient market where A = A = · · · = A q = 0 , the degree ζ becomes zero; otherwise, ζ deviates from zero. Thus, we call ζ the joint degree of market efficiency. When we finda large deviation of ζ from (both positive and negative), we can regard some deviationfrom one as evidence of market inefficiency. Furthermore, we can construct this degreethat would vary over time when we obtain time-varying estimates of the coefficients inEquation (3).We estimate TV-VAR coefficients in each period to obtain the degree defined in Equa-tion (5) in each period. In practice, we employ a model in which all the VAR coefficients,except for the one that corresponds to intercept terms, ν , follow independent randomwalk processes. In other words, we assume: A l,t = A l,t − + V l,t , ( l = 1 , , · · · , q ) , (6)where { V l,t } is a q × t dimensional error term matrix. We assume that the matrix satisfies E [ V l,t ] = O for all t , E [ vec ( V l,t ) (cid:48) vec ( V l,t )] = σ v I and E [ vec ( V l,t ) (cid:48) vec ( V l,t − m )] = O for all l and m (cid:54) = 0 . Ito et al.’s (2014; 2017) method allows us to estimate the TV-VAR model: x t = ν + A ,t x t − + A ,t x t − + · · · + A q,t x t − q + ε t , (7)together with Equation (6).To conduct statistical inference on a joint degree of market efficiency in the semi-strongsense, we apply a residual bootstrap technique to the TV-VAR model above. In practice,we build a set of bootstrap samples of the TV-VAR estimates under the hypothesis thatall the TV-VAR coefficients are zero. This procedure provides us with a (simulated)distribution of the estimated TV-VAR coefficients, assuming the three securities return6rocesses are generated under the semi-strong form EMH. Thus, we can compute thecorresponding distribution of the impulse response and degree of market efficiency inthe semi-strong sense. Finally, by using confidence bands derived from such simulateddistribution, we conduct statistical inference on our estimates and detect periods whenthe prewar Japanese stock exchange experienced market inefficiency. This paper studies the dynamics of the semi-strong efficiency of the prewar Japanese stockmarket, investigating how the stock prices (or total returns) responded to policy changesand external events. That is, we consider not only the stock but also the governmentbond and the foreign exchange when we measure the time-varying joint degree of marketefficiency in the prewar Japanese stock market. As stock prices, we utilize the three typesof total return indices that are calculated by Hirayama (2017a,b, 2018a, 2019, 2020a):the old shares of the Tokyo Stock Exchange (hereafter the old shares), the new sharesof the Tokyo Stock Exchange (hereafter the new shares), and the equity performanceindex (EQPI). Note that the EQPI is the capitalization-weighted index while the oldand new shares are based on the stock price of the Tokyo Stock Exchange (TSE) andboth are price-weighted indices. Then the government bond performance index (GBPI) ofHirayama (2018b, 2020b) is used as the government bond prices. Hirayama calculates thetotal return for the prewar Japanese government bond and constructed GBPI as a newgovernment bond price index. Lastly, we obtain current exchange rates of the U.S. dollaragainst Japanese yen from the
Bulletin of Financial Information (Ministry of Finance,Japan) and Bank of Japan (1947) as the foreign exchange prices. We then construct three datasets for each stock price index. The start periods are thesame regardless of datasets, June 1924. However, the end periods depend on the dataset;April 1943 for the old and new shares, and August 1945 for the EQPI. The sample periodsof the GBPI and the exchange rate are both from June 1924 to August 1945. We takethe log first difference of the time series of prices to obtain the returns of the indices.(Table 1 around here)Table 1 shows the descriptive statistics for the returns. We confirm that the mean (stan-dard deviation) of returns on the EQPI is higher (lower) than that of the old and thenew shares price index. This means that the differences in the weighting method forthe price index affect the performance of the stock. We also confirm that the standarddeviation of the returns on GBPI is the lowest in all of the securities. This implies thatthe government bond was considered a relatively low-risk asset in prewar Japan.Table 1 also shows the results of the unit root test with descriptive statistics forthe data. For estimations, all variables that appear in the moment conditions shouldbe stationary. We apply the Elliott et al.’s (1996) augmented Dickey–Fuller generalizedleast squares (ADF-GLS) test to confirm whether the variables satisfy the stationaritycondition. We employ the modified Bayesian information criterion (MBIC) instead ofthe modified Akaike information criterion (MAIC) to select the optimal lag length. Thisis because from the estimated coefficient of the detrended series, ˆ ψ , we do not find thepossibility of size-distortions (see Elliott et al. (1996); Ng and Perron (2001)). The ADF- Please see the appendix at the end of this paper on how the dataset is constructed.
We first assume a time-invariant VAR( q ) model with constant parameters and employSchwarz’s (1978) Bayesian information criteria to select the optimal lag order in ourpreliminary estimations. Table 1 summarizes our preliminary results for a time-invariantVAR( q ) model using the whole sample. In the estimations, we choose first-order vectorautoregressive (VAR( )) models for all estimations.(Table 2 around here)Table 2 shows that (1) the VAR estimates of the old and new shares are almost similar,and (2) the estimates of R S,t − of the EQPI is smaller than those of the old and newshares. This means that when we consider the semi-strong form market efficiency, theEQPI is the most efficient in the prewar Japanese stock market. Note that because ofthis result as well as the limited explanatory power of time-invariant VAR models, weshould pay more attention to the time-varying nature of the semi-strong market efficiencyon the prewar Japanese stock market.We employ Hansen’s (1992) parameter constancy test under the random parametershypothesis to detect whether the parameters are constant in the above VAR( q ) models.Table 2 also presents the test statistics; we reject the null of constant parameters againstthe parameter variation as a random walk at the 5% significance level. Therefore, weestimate the time-varying parameters of the above VAR models to investigate whethergradual or rapid changes occur in the prewar Japanese stock market. These resultssuggest that the time-invariant VAR( q ) model is not suitable to explain our data andthat the TV-VAR( q ) model is a better fit.From a historical viewpoint, the prewar Japanese economy experienced various ex-ogenous shocks such as bubbles, economic or political crises, policy changes, and wars.Table 3 summarizes the major historical events in the prewar Japanese economy.(Table 3 around here)We consider that the events affected the stock price formation. Accordingly, we estimatethe joint degree of market efficiency using the GLS-based TV-VAR model in the nextsubsection. (Figure 1 around here)Figure 1 shows that the joint degree of market efficiency in the prewar Japanese stockmarket fluctuated over time. We confirm that the variation of the degrees looks similar Note that are some related studies about examining the weak-form EMH in prewar Japan such asIto et al. (2016, 2018). They estimate the time-varying degree of market efficiency in the prewar Japaneserice markets and find that the market efficiency changes through time. After 1926, however, the rise in stock prices ended as the yen appreciated against theU.S. dollar due to heightened speculation about the lifting of the gold embargo. Hence,the Japanese stock market seems to have faded away from expectations of a strong yenand economic austerity, which had been fueled by speculation about the lifting of the goldembargo. The total return index of the EQPI also rose until February 1926. However, itcan be confirmed that the rate of increase of the old and new shares is much higher thanthat of EQPI.In practice, the total return index of the EQPI increased by 12.0% between Decem-ber 1925 and February 1926, while the total returns of TSE (spot transaction) and thenew TSE (short-term clearing transaction) increased by 37.6% and 36.8%, respectively,showing that the stock investment performance of the TSE has been overwhelmingly en-hanced. This may be because the additional payment of the new shares was decided onJanuary 8, 1926 (the payment date is April 1, 1926), and the stock price was expectedto rise. The additional payment of new shares was seen as an indication of the firm’sstrong demand for funds despite the increased burden on shareholders. Therefore, ex-pectations for a rise in stock prices increased and speculative money flowed into the newshares of the TSE. The new shares of the TSE may have been overreacting, partly due tothe increase in trading volume on the exchange. For example, when the TSE decided tomake additional payments in December 1931, in a case, the stock price of the TSE washigher than that of the EQPI due to an increase in speculative interest. This may suggestthe following two points. The first point is that changes in the efficiency of individualstocks are affected by the impact of additional payments. In the efficiency test, studiesthat rely on individual stock data are not appropriate because they include this noise.Therefore, when examining the efficiency of the prewar stock market, an index that in-cludes a large number of stocks (all stocks if possible) should be used. The second pointis that even an index containing a large number of stocks has a phase in which efficiencydeclines (June 1932), suggesting that efficiency for the entire market has changed. A testusing an appropriate stock index after making various adjustments would contribute toa sophisticated analysis of prewar stock market efficiency.During the second half of 1928, the old and new shares of TSE rallied mainly onthe back of low interest rates and monetary easing (see Tokyo Stock Exchange (1933,pp.39–40) for details), but in May 1929, the stock market was upset by rumors that thegovernment was likely to lift the gold embargo. As the yen continued to appreciatetoward the end of the year, the stock market plummeted, led by the old and new sharesof the TSE. The difference between the old and new shares of the TSE is the variation ofdegrees after the Pacific War occurred in December 1941. The degree of the new sharesrapidly declined, but the degree of the old TSE did not. When stock prices were expectedto rise,
Oyafuko-Souba (lack of filial piety markets; new share is more expensive than oldshare or child share is more expensive than parent share) frequently appeared on thestock market before WWII. For most of the period since March 1932, the old and newshares of the TSE shares have been in
Oyafuko-Souba . Oyafuko-Souba means that the See Tokyo Stock Exchange (1974, pp.42–43) for details. See Fukai (1941, p.236) for details.
Oyafuko-Souba is the inflow of speculative money due to the expectation of a rise in stock prices.In December 1941, the regulatory authority for stock exchanges was transferred from theMinistry of Commerce and Industry to the Ministry of Finance. On December 26, 1942,it was decided that stock exchanges would be delisted. Since January 1943, it is clearlyno longer
Oyafuko-Souba as the new shares has declined significantly compared to theold shares. Considering the delisting decision, the premium of the new shares, which hadbeen positioned as a highly liquid benchmark, rapidly eroded. This significant drop inthe new shares may have been a factor in increasing inefficiency.On the other hand, the variation of the degree in the case of the EQPI is quite differentfrom those of the old and new shares. The degree hardly fluctuated except for the periodof the establishment of the foreign exchange control law in June 1932. In summary,we conclude that market efficiency varied over time because of external events such aspolicy changes and wars. These results are consistent with the results of Noda (2019)who measures the degree of market efficiency in the sense of weak-form using the prewarJapanese stock market data.
Next, we examine the time-varying impulse responses to investigate how externalevents as shocks affect the three indices.(Figures 2, 3, and 4 around here)Figures 2, 3, and 4 show the time-varying impulse responses for each security. The non-diagonal figures indicate that the shock of the foreign exchange rate is the biggest, followedby those of the government bond and stock. Particularly, it seems that the establishmentof the foreign exchange control law in June 1932 affected the stock and the governmentbond through the foreign exchange rate. However, the outbreak of the Pacific War inDecember 1941 appears to have little effect on the securities. To explore in detail theimpact of external events, we then confirm the static impulse response in certain periodsby splitting the above time-varying impulse responses. Particularly, we pay attention tothe three periods: (1) the abandonment of the gold standard in December 1931, (2) theestablishment of the foreign exchange control law in June 1932, and (3) the Pacific Waroccurred on December 1941.(Figures 5, 6, and 7 around here)Figures 5, 6, and 7 show the static impulse response at the above three periods. We cansee that the impulse responses in the case of the outbreak of the Pacific War decay morerapidly than those of the abandonment of the gold standard and the establishment of theforeign exchange control law. It means that the markets rapidly reflect the informationof the external events with time. Moreover, we confirm that the instantaneous impulseresponse of the outbreak of the Pacific War is incomparable with those of policy changesrespect with the foreign exchange.As pointed out by Ito (1989, pp.271–278), in spite of the decision to ban gold exportsagain in December 1931 and the enactment of the Act on the Prevention of Fund Evasionin July 1932, there was a time lag before strict implementation of exchange rate controls.10articularly, the “Chinese Speculator” speculatively traded Japanese yen in the Shanghaiand Dalian markets, accelerating the appreciation of the yen. It can be said that therewas a strong sense of uncertainty about the government’s exchange rate policy untilthe Foreign Exchange Control Law was enacted in 1933. Subsequently, the price keepingoperation by the Wartime Finance Bank started in Summer 1944, and in earnest in March1945. At wartime, Hirayama (2018b) confirms the difference in the level of volatilitybetween stock market and government bond and foreign exchange markets. In contrast,the stock market during the war had, to some extent, a market pricing mechanism.The price keeping operation by the Wartime Finance Bank started in earnest in March1945. Hirayama (2018a) confirms the difference in the level of volatility. Therefore, thesensitivity of the stock market to external information was maintained to some extentcompared with the government bond and foreign exchange markets.
This paper examines Fama’s (1970) semi-strong form EMH in the prewar Japanese stockmarket using a new dataset obtained from Hirayama (2017a,b, 2018a,b, 2019, 2020a,b).Particularly, we focus on the relationship between the prewar Japanese stock market andseveral government policy interventions to explore whether the semi-strong form stockmarket efficiency evolves over time. We consider that government policy interventionschange the semi-strong stock market efficiency through the distortion of security pricessuch as stock, government bonds, and foreign exchange. To capture the long-run impactof government policy interventions against stock markets, we measure the time-varyingjoint degree of market efficiency and the time-varying impulse responses based on Itoet al.’s (2014; 2017) GLS-based TV-VAR model instead of the conventional approachsuch as the event study analysis used in most of the previous studies. We summarize theresults as follows. First, the joint degree of market efficiency in the prewar Japanese stockmarket fluctuated over time because of external events such as policy changes and wars.These results are consistent with the results of Noda (2019) who measures the degree ofmarket efficiency in the sense of weak-form using the prewar Japanese stock market data.Second, the semi-strong form EMH is almost supported in the prewar Japanese stockmarket as well as the results of Suzuki and Yuki (2019). They show that stock pricesreflected various data correctly even after the Kanto Great Earthquake occurred; and theyconclude the semi-strong form EMH is supported. Third, we can see that the impulseresponses in the case of the outbreak of the Pacific War decay more rapidly than thoseof the abandonment of the gold standard and the establishment of the foreign exchangecontrol law. It means that the markets now reflect the information of the external eventsto the stock prices faster than before. Therefore, we conclude that Lo’s (2004) AMH isalso supported in the prewar Japanese stock market even if we consider that the publicinformation affects the stock market. 11 cknowledgments
The authors would like to thank Mikio Ito, Yumiko Miwa, Masato Shizume, Shiba Suzuki,Tatsuma Wada, Takenobu Yuki, and the seminar participants at Keio University for theirhelpful comments and suggestions. The author (Noda) is also grateful for the financialassistance provided by the Japan Society for the Promotion of Science Grant in Aid forScientific Research, under grant numbers 17K03809, 18K01734, and 19K13747. All dataand programs used are available upon request.
References
Ball, R. and Brown, P. (1968), “An Empirical Evaluation of Accounting Income Numbers,”
Journal of Accounting Research , 6, 159–178.Bank of Japan (1947), “Senji Chu Tokei Yoran [Handbook of Statistics During the War],”Tokyo, Japan.Bassino, J. and Lagoarde-Segot, T. (2015), “Informational Efficiency in the Tokyo StockExchange, 1931–40,”
Economic History Review , 68, 1226–1249.Becker, K. G., Finnerty, J. E., and Kopecky, K. J. (1996), “Macroeconomic News andthe Efficiency of International Bond Futures Markets,”
Journal of Futures Markets , 16,131–145.Bernanke, B. B. (2004), “What Policymakers can Learn from Asset Prices,” Tech. rep.,Remarks before The InvestmentAnalysts Society of Chicago.Brown, S. J. and Warner, J. B. (1980), “Measuring Security Price Performance,”
Journalof Financial Economics , 8, 205–258.Campbell, J. Y. (1987), “Stock Returns and the Term Structure,”
Journal of FinancialEconomics , 18, 373–399.Campbell, J. Y. and Shiller, R. J. (1988a), “The Dividend-Price Ratio and Expectationsof Future Dividends and Discount Factors,”
Review of Financial Studies , 1, 195–228.— (1988b), “Stock Prices, Earnings, and Expected Dividends,”
Journal of Finance, , 43,661–676.Choudhry, T. (2010), “World War II Events and the Dow Jones Industrial Index,”
Journalof Banking and Finance , 34, 1022–1031.Chuliá, H., Martens, M., and van Dijk, D. (2010), “Asymmetric Effects of Federal FundsTarget Rate Changes on S&P100 Stock Returns, Volatilities and Correlations,”
Journalof Banking & Finance , 34, 834–839.Darrat, A. F. (1988), “On Fiscal Policy and the Stock Market,”
Journal of Money, Creditand Banking , 20, 353–363.Davidson, L. S. and Froyen, R. T. (1982), “Monetary Policy and Stock Returns:Are StockMarkets Efficient?”
Federal Reserve Bank of St. Louis Review , 64, 3–12.12hrmann, M. and Fratzscher, M. (2004), “Taking Stock: Monetary Policy Transmissionto Equity Markets,”
Journal of Money, Credit and Banking , 36, 719–737.Elliott, G., Rothenberg, T. J., and Stock, J. H. (1996), “Efficient Tests for an Autore-gressive Unit Root,”
Econometrica , 64, 813–836.Fama, E. F. (1970), “Efficient Capital Markets: A Review of Theory and Empirical Work,”
Journal of Finance , 25, 383–417.— (1981), “Stock Returns, Real Activity, Inflation, and Money,”
American EconomicReview , 71, 545–565.— (1991), “Efficient Capital Markets: II,”
Journal of Finance , 46, 1575–1617.Fama, E. F., Fisher, L., Jensen, M. C., and Roll, R. (1969), “The Adjustment of StockPrices to New Information,”
International Economic Review , 10, 1–21.Fama, E. F. and French, K. R. (1988), “Dividend Yields and Expected Stock Returns,”
Journal of Financial Economics , 22, 3–25.— (1989), “Business Conditions and Expected Returns on Stocks and Bonds,”
Journal ofFinancial Economics , 25, 23–49.Fujino, S. and Teranishi, J. (2000),
A Quantitative Analysis of the Japanese FinancialSystem , Toyo Keizai Shimpo Sha.Fukai, E. (1941),
Kaiko 70-nen [Reflections of 70 Years], Iwanami Shoten, Tokyo.Fukami, Y. (2012), “History of Institute for Stock Price- Keeping in Prewar Showa PeriodFocusing on the Seiho-Shouken (in Japanese),”
Journal of Financial and SecuritiesMarkets (Japan Securities Research Institute), 78, 1–18.Hancock, D. G. (1989), “Fiscal Policy, Monetary Policy and the Efficiency of the StockMarket,”
Economics Letters , 31, 65–69.Hansen, B. E. (1992), “Testing for Parameter Instability in Linear Models,”
Journal ofPolicy Modeling , 14, 517–533.Harada, K. and Okimoto, T. (2019), “The BOJ’s ETF Purchases and Its Effects on Nikkei225 Stocks,” RIETI Discussion Paper Series 19-E-014.Hirayama, K. (2017a), “The Japanese Stock Market Performance Index in the EarlyShowa Era (in Japanese),”
Annals of Society for the Economic Studies of Securities (Japan Securities Research Institute), 51, 1–12.— (2017b), “Reappraisal of Japanese Equity Market Return in the Early Showa Era (inJapanese),”
Journal of Economic Science (The Economics Society of Saitama Univer-sity), 14, 41–53.— (2018a), “The Japanese Equity Performance Index in the Early Showa Era (inJapanese),”
Journal of Financial and Securities Markets (Japan Securities ResearchInstitute), 101, 71–91. 13 (2018b), “The Japanese Government Bond Performance Index in the Early Showa Era(in Japanese),”
Review of Monetary and Financial Studies (Japan Society of MonetaryEconomics), 40, 54–65.— (2019), “A Linkage between Prewar and Postwar Price in the Japanese Stock Market,”Institute for Stock Price Index Workshop at Meiji University (August 19, 2019).— (2020a), “The Japanese Equity Performance from 1944 to 1945 (in Japanese),”
Journalof Financial and Securities Markets (Japan Securities Research Institute), 109, 63–85.— (2020b), “The Japanese Government Bond Performance from 1944 to 1945 (inJapanese),” Mimeo.Institute for Monetary and Economic Studies, Bank of Japan (1993), “Chronology of Fi-nancial Matters in Japan (in Japanese),” Institute for Monetary and Economic Studies,Bank of Japan, revised Edition.Ito, M. (1989),
Nihon no Taigai Kin’yu to Kin’yu Seisaku (Japan’s Overseas Financesand Monetary Policies: 1914-1936), Nihon Keizai Hyoron Sha, Tokyo.Ito, M., Maeda, K., and Noda, A. (2016), “Market Efficiency and Government Inter-ventions in Prewar Japanese Rice Futures Markets,”
Financial History Review , 23,325–346.— (2018), “The Futures Premium and Rice Market Efficiency in Prewar Japan,”
EconomicHistory Review , 71, 909–937.Ito, M., Noda, A., and Wada, T. (2014), “International Stock Market Efficiency: A Non-Bayesian Time-Varying Model Approach,”
Applied Economics , 46, 2744–2754.— (2017), “An Alternative Estimation Method of a Time-Varying Parameter Model,”[arXiv:1707.06837], Available at https://arxiv.org/pdf/1707.06837.pdf.Ito, O. (1999), “The Transformation of the Japanese Economy,” in
Japan’s War Econ-omy (Routledge Studies in the Growth Economies of Asia), ed. Pauer, E., Routledge,London, chap. 9, pp. 171–187.Kasuya, M. (2006), “Securities Investments of Regional Banks during Interwar PeriodJapan,”
Monetary and Economic Studies , 25, 59–104.Kataoka, Y., Maru, J., and Teranishi, J. (2004), “An Analysis of Stock Market Efficiencyin the Late Meiji Era, Part.2 (in Japanese),”
Journal of Financial and Securities Mar-kets (Japan Securities Research Institute), 48, 69–81.Kato, T. (1965a), “Banking Conditions under the War Economy : Through an Analysisof Big Banks (I),”
Journal of Social Science , 17, 1–42.— (1965b), “Banking Conditions under the War Economy : Through an Analysis of BigBanks (I),”
Journal of Social Science , 17, 94–119.Kobayashi, K. (2012),
Nihon Shoken Shiron (History of Japanese Securities), Nihon KeizaiHyoron Sha, Tokyo. 14aopodis, N. T. (2009), “Fiscal Policy and Stock Market Efficiency: Evidence for theUnited States,”
Quarterly Review of Economics and Finance , 49, 633–650.Lim, K. P. and Brooks, R. (2011), “The Evolution of Stock Market Efficiency Over Time:A Survey of the Empirical Literature,”
Journal of Economic Surveys , 25, 69–108.Lo, A. W. (2004), “The Adaptive Markets Hypothesis: Market Efficiency from an Evolu-tionary Perspective,”
Journal of Portfolio Management , 30, 15–29.MacKinlay, A. C. (1997), “Event Studies in Economics and Finance,”
Journal of EconomicLiterature , 35, 13–39.Malkiel, B. G. (2003), “The Efficient Market Hypothesis and its Critics,”
Journal ofEconomic Perspectives , 17, 59–82.McWilliams, A. and Siegel, D. (1997), “Event Studies in Management Research: Theo-retical and Empirical Issues,”
Academy of Management Journal , 40, 626–657.Newey, W. K. and West, K. D. (1987), “A Simple, Positive Semi-Definite, Heteroskedas-ticity and Autocorrelation Consistent Covariance Matrix,”
Econometrica , 55, 703–708.Ng, S. and Perron, P. (2001), “Lag Length Selection and the Construction of Unit RootTests with Good Size and Power,”
Econometrica , 69, 1519–1554.Noda, A. (2019), “Measuring the Time-Varying Market Efficiencyin the Prewar JapaneseStock Market,” [arXiv:1911.04059], Available at https://arxiv.org/pdf/1911.04059.pdf.Pearce, D. and Roley, V. (1985), “Stock Prices and Economic News,”
Journal of Business ,58, 49–67.Perron, P. (1989), “The Great Crash, the Oil Price Shock, and the Unit Root Hypothesis,”
Econometrica , 57, 1361–1401.Schwarz, G. (1978), “Estimating the Dimension of a Model,”
Annals of Statistics , 6,461–464.Shibamoto, M. and Shizume, M. (2014), “Exchange Rate Adjustment, Monetary Policyand Fiscal Stimulus in Japan’s Escape from the Great Depression,”
Explorations inEconomic History , 53, 1–18.Shibata, Y. (2011),
Senji Nihon no Kinyu Tosei [Monetary Controls in Wartime Japan],Nihon Keizai Hyoron Sha, Tokyo.Suzuki, S. (2012), “Pacific War and Tokyo Stock Exchange Daily Data:1941-1943 (inJapanese),”
Bulletin of Economic Studies (Meisei University), 44, 39–51.Suzuki, S. and Yuki, T. (2019), “Great Kanto Earthquake and the Japanese Stock Market(in Japanese),” Mimeo.Tokyo Stock Exchange (1928),
Fifty Years History of the Tokyo Stock Exchange (inJapanese), Tokyo Stock Exchange.— (1933),
History of the Tokyo Stock Exchange (in Japanese), vol. 2, Tokyo Stock Ex-change. 15 (1974),
History of the Tokyo Stock Exchange Twenry Years (in Japanese), Tokyo StockExchange.Ueda, K. (2012), “The Effectiveness of Non-traditional Monetary Policy Measures: TheCase of the Bank of Japan,”
Japanese Economic Review , 63, 1–22.— (2013), “Response of Asset Prices to Monetary Policy under Abenomics,”
Asian Eco-nomic Policy Review , 8, 252–269.Utsunomiya, K. (2013), “Japan’s Financial Intermediation in the 1940s: Estimation ofFlow of Funds Accounts from 1941 to 1948 (in Japanese),”
Review of Monetary andFinancial Studies (Japan Society of Monetary Economics), 35, 52–73.16 ppendix
A.1 Dataset on the Equity Markets
In prewar Japan, the system of payment of stock in installations was adopted forthe issue of stocks of companies. This was a capital system in which shareholders paidthe par value of their shares in installments rather than in full at one time. Hence, theburden on shareholders to make payments was reduced and smooth capital concentrationwas promoted. In principle, a new capital increase by issuing new shares is possible oncethe entire par value (most of them are 50 yen) is paid in. The former shares were calledold shares (a fully paid stock) and the latter shares were called new shares (a part-paidstock). Therefore, one company issued several types of stocks; and at the same time,they were traded as different stocks. As for shares issued by the Tokyo Stock Exchange(TSE), which had a large volume of trading, old shares with the face value fully paid andnew shares that were unpaid existed and were simultaneously listed on the exchange .Hereinafter, we will be refer to them as the old and the new shares, respectively.The system of payment of stock in installations has made the calculation of stockinvestment performance more complicated by factors such as allotment of new sharesto shareholders and additional payments. Therefore, it is difficult to simply comparethe results of stock investment in the prewar and wartime periods with those of modernstock investment. It is necessary to calculate accurate investment results by making threeadjustments: correction of ex-right, correction of additional payments, and correction ofdividends. From this viewpoint, three adjustments were made to two types of stocks ofthe TSE, which had been treated as representative stocks on the stock market from theMeiji period to the prewar and wartime periods, to calculate accurate investment results.Because of these adjustments, three types of indices were calculated: a price index (PI)linked to the stock price, an adjusted price index (API) with an adjustment of the ex-rights (old shares) and additional payments (new shares), and a total return index (TRI)with an adjustment of the dividend rights.However, while it has the advantage of being able to calculate long-term data andbeing an indicator of the degree of activity in the stock market, it is difficult to reflectthe performance of the heavy and chemical industries, which rapidly expanded duringwartime. Hence, it cannot be considered as a suitable indicator of the stock prices ofmajor Japanese companies. Then Hirayama (2017a,b, 2018a, 2019, 2020a) provide themonthly equity performance index (EQPI) from June 1924 to August 1945. The EQPIis a market index of listed stocks on the TSE’s short-term transactions (a kind of futurequotations) in stocks that is weighted by market capitalization (26 issues were listed atthe same time, for a total of 38 issues). Since September 1943, the same issues have Note that Japanese stock exchanges before the war were listed companies until 1943. The shares (a fully-paid stock) represent the following 24 issues: Dalian Stock Commodity Ex-change, Manchuria Heavy Industries, Nippon Oil, Ensuiko Sugar Refining, Tokyo Electric Light, NichiroFisheries, Kanegafuchi Spinning, Meiji Sugar Manufacturing, Oji Paper, Nisshin Cotton Spinning, Mit-subishi Mining, Nippon Kokan, Nippon Electric Power, Hokkaido Colliery & Steamship, DainipponJinzo Hiryo, Nippon Yusen Kaisha, Mitsubishi Heavy Industries, Showa Fertilizer, Nippon Mining, Hi-tachi Seisakusho, Nippon Electrical Industries, Nippon Soda, Kokura Steel Manufacturing, and RasaIndustries. The new shares (a part-paid stock) represent the following 14 issues: Tokyo Stock Exchange(New), Kanegafuchi Spinning (New), Nippon Yusen Kaisha (New), Dainippon Sugar Manufacturing(New), Asano Cement(New), Fuji Paper(New), Dainippon Beer(New), South Manchuria Railway (New),South Manchuria Railway (Second New), Osaka Stock Exchange (New), Teikoku Jinzo-Kenshi Kaisha(New), Toyo Rayon (New), Nippon Mining (New), and Nippon Suisan (New).
Monthly Statistical Reports (Tokyo/Japan Stock Exchange), and from December 1944 onward, the
Nihon SangyoKeizai (forerunner of the Nihon Keizai Shimbun).
A.2 Dataset on the Government Bonds Market
In the past, the yields on the 5% Loan (Mark Ko) and 4% Loan (1st issue) were usedas indicators of the prewar and wartime government bond markets. However, analysis ofthese individual issues does not show actual investment performance in the prewar andwartime Japanese government bond (JGB) markets, because yield spreads exist due todifferences in coupon rates. Hirayama (2018b, 2020b) provides a capitalization-weightedindex for all interest-bearing government bonds with a maturity of at least one yearand calculates the monthly government bond performance index (GBPI) for the periodfrom June 1924 to August 1945 in accordance with the standard methods of modernbond indices. There are two types of GBPI: the PI and the TRI that also reflects incomereturns. Domestic government bonds for the period for index calculation can be classifiedinto 20 types, and the total number of issues is 268. Hirayama refers to the followingresources for calculating the GBPI: the JGB price quotes specified in the monthly averageprices in the Tokyo market stated in the
Debt Management Reports (Ministry of Finance,Japan) from June 1924 to March 1942, and the spot transaction and monthly averageprice stated in the
Monthly Statistical Reports (Tokyo/Japan Stock Exchange) from April1942 to November 1944. Because of tighter price controls after December 1944, the bondprices were fixed, and November 1944 values were adopted.
A.3 Dataset on the Exchange Rate
Data before 1942 are official prices listed in
Bulletin of Financial Information (Min-istry of Finance, Japan) for each fiscal year, and those after 1942 are official prices listedin Bank of Japan (1947). Indexes are calculated for 5% Loan, 5% Loan (Special), 5% Loan (Mark Ko), 4% Loan (1st issue),4% Loan (2nd issue), 4% Loan, 3.5% Loan, 5% Treasury Bonds, Railway Bonds, Extraordinary TreasuryBonds (Issued at Discount), 4.5% Treasury Bonds, 4% Treasury Bonds, 3.5% Treasury Bonds, ChinaIncident (Second Sino-Japanese War) Treasury Bonds, China Incident Treasury Bonds (Special), TheGreat East Asia War (Pacific War) Treasury Bonds, The Great East Asia War (Pacific War) TreasuryBonds (Special), and Grants Treasury Bonds Delivered. − . − . − . − . − . Max 0.2209 0.2258 0.1991 0.0293 0.3229ADF-GLS − . − . − . − . − . Lags 0 0 0 0 0 ˆ φ N
226 226 245 245 245
Notes:(1) “Old,” “New,” and “EQPI” denote the old shares of the Tokyo Stock Exchange, the newshares of the Tokyo Stock Exchange, and the equity performance index, respectively.(2) “GBPI” and “Exchange” denote the returns on the government bond performanceindex and the exchange rate, respectively.(3) “ADF-GLS” denotes the ADF-GLS test statistics, “Lags” denotes the lag order selectedby the MBIC, and “ ˆ φ ” denotes the coefficients vector in the GLS detrended series (seeEquation (6) in Ng and Perron (2001)).(4) In computing the ADF-GLS test, a model with a time trend and a constant is assumed.The critical value at the 5% significance level for the ADF-GLS test is “ − . ”.(5) “ N ” denotes the number of observations.(6) R version 4.0.2 was used to compute the statistics. a b l e : P r e li m i n a r y E s t i m a t i o n s a nd P a r a m e t e r C o n s t a n c y T e s t s O l dSh a r e s N e w Sh a r e s E Q P I S t o c k G B P I E x c h a n g e S t o c k G B P I E x c h a n g e S t o c k G B P I E x c h a n g e C o n s t a n t . . . . . . . . . [ . ][ . ][ . ][ . ][ . ][ . ][ . ][ . ][ . ] R S , t − . − . . . − . . . − . . [ . ][ . ][ . ][ . ][ . ][ . ][ . ][ . ][ . ] R G , t − − . . . − . . . . . . [ . ][ . ][ . ][ . ][ . ][ . ][ . ][ . ][ . ] R E , t − − . − . . − . − . . . − . . [ . ][ . ][ . ][ . ][ . ][ . ][ . ][ . ][ . ] ¯ R . . . . . . . . . L C . . . N o t e s : ( ) “ R t − p , ”“ ¯ R , ” a nd “ L C ” d e n o t e t h e VA R ( p ) e s t i m a t e , t h e a d j u s t e d R , a nd t h e H a n s e n ’ s ( ) j o i n t L s t a t i s t i c w i t h v a r i a n ce , r e s p ec t i v e l y . ( ) N e w e y a nd W e s t ’ s ( ) r o bu s t s t a nd a r d e rr o r s a r e b e t w ee nb r a c k e t s . ( ) R v e r s i o n . . w a s u s e d t o c o m pu t e t h ee s t i m a t e s . Periods Major historical eventsJanuary 1926 TSE’s new share price skyrocketed with the announcement of ad-ditional payment (from 12.50 yen to 25.00 yen).April 1927 The nationwide financial panic sparked when debates in the Dietrevealed financial difficulties between the Bank of Taiwan andSuzuki & Co. (The Showa Financial Crisis); The amended Bank-ing Law promulgated on March 30.July 1929 The Hamaguchi Cabinet announced the time for the lifting of anembargo on the export of gold.October 1929 The Great Crash occurred.November 1930 Seiho-Shouken (Stock Price Keeping Organization cooperated byJapan’s Life Insurance Companies) established.January 1931 The Japanese government lifted an embargo on the export of gold.September 1931 The U.K. abolished the gold standard on 21 September; With therise of dollar buying speculation in the foreign exchange market,the value of government bonds and stocks declined.November 1931 Expectations for a re-ban on gold exports rose, and stock pricesstarted to reverse and rise.December 1931 The Inukai Cabinet abandoned the gold standard on 13 December;The TSE’s new share price skyrocketed with the announcement ofadditional payment (from 25.00 yen to 37.50 yen). In contrast toJapan, stock prices in the U.S., U.K., and France were declining.January 1932 The official exchange rate of the Japanese yen against the U.S.dollar rapidly decreased.July 1932 The Capital Flight Prevention Law and The Act on the Calcula-tion of Government Bond Prices promulgated and enforced on 1JulyNovember 1932 The Bank of Japan started underwriting government bonds.May 1933 The Foreign Exchange Control Law enforced.February 1936 February 26 incident occurred; Korekiyo Takahashi’s economicpolicy ended.July – August 1937 The Second Sino-Japanese War occurred.July 1941 President Franklin Roosevelt froze all Japanese assets in the U.S.December 1941 The Pacific War occurred; The Ministry of Finance transferred theadministrative work of the exchange to the Ministry of Commerceand Industry.June 1942 The Battle of Midway occurred.December 1942 The Japanese government decided to delist the TSE shares.February 1943 The Operation Ke completed (withdrawal of Japanese forces fromGuadalcanal).July 1944 The Operation Forager (the Battle of Saipan) completed.July/August 1944 The Wartime Finance Bank started the stock price-keeping oper-ation.March 1945 The Great Tokyo Air Raids occurred; The Wartime Finance Bankstarted the stock price-keeping operation (unlimited stock pur-chase). In July, the Japan stock exchange started the stock price-keeping operation.Note: This table is constructed following Institute for Monetary and Economic Studies, Bank ofJapan (1993) and Fukami (2012). Old Shares
Year D eg r ee o f M a r k e t E ff i c i en cy . . . . . . . . . . . . . . . . . . . . . New Shares
Year D eg r ee o f M a r k e t E ff i c i en cy . . . . . . . . . . . . . . . . . . . . . EQPI
Year D eg r ee o f M a r k e t E ff i c i en cy . . . . . . . . . . . . . . . . . . . . . Notes:(1) The dashed red lines represent the 95% confidence intervals of the efficientmarket degrees.(2) We run the bootstrap sampling 5,000 times to calculate the confidence inter-vals.(3) R version 4.0.2 was used to compute the estimates.22igure 2: Time-Varying Impulse Responses (Old Shares)
Lag s T i m e I m pu l s e R e s pon s e Old Shares to Old Shares
Lag s T i m e I m pu l s e R e s pon s e Old Shares to GBPI
Lag s T i m e I m pu l s e R e s pon s e Old Shares to Exchange
Lag s T i m e I m pu l s e R e s pon s e GBPI to Old Shares
Lag s T i m e I m pu l s e R e s pon s e GBPI to GBPI
Lag s T i m e I m pu l s e R e s pon s e GBPI to Exchange
Lag s T i m e I m pu l s e R e s pon s e Exchange to Old Shares
Lag s T i m e I m pu l s e R e s pon s e Exchange to GBPI
Lag s T i m e I m pu l s e R e s pon s e Exchange to Exchange
Note: R version 4.0.2 was used to compute the estimates.23igure 3: Time-Varying Impulse Responses (New Shares)
Lag s T i m e I m pu l s e R e s pon s e New Share to New Share
Lag s T i m e I m pu l s e R e s pon s e New Share to GBPI
Lag s T i m e I m pu l s e R e s pon s e New Share to Exchange
Lag s T i m e I m pu l s e R e s pon s e GBPI to New Share
Lag s T i m e I m pu l s e R e s pon s e GBPI to GBPI
Lag s T i m e I m pu l s e R e s pon s e GBPI to Exchange
Lag s T i m e I m pu l s e R e s pon s e Exchange to NEW SHARES
Lag s T i m e I m pu l s e R e s pon s e Exchange to GBPI
Lag s T i m e I m pu l s e R e s pon s e Exchange to Exchange
Note: As for Figure 2. 24igure 4: Time-Varying Impulses Response (EQPI)
Lag s T i m e I m pu l s e R e s pon s e EQPI to EQPI
Lag s T i m e I m pu l s e R e s pon s e EQPI to GBPI
Lag s T i m e I m pu l s e R e s pon s e EQPI to Exchange
Lag s T i m e I m pu l s e R e s pon s e GBPI to EQPI
Lag s T i m e I m pu l s e R e s pon s e GBPI to GBPI
Lag s T i m e I m pu l s e R e s pon s e −0.0050.0000.0050.0100.0150.020 GBPI to Exchange
Lag s T i m e I m pu l s e R e s pon s e Exchange to EQPI
Lag s T i m e I m pu l s e R e s pon s e Exchange to GBPI
Lag s T i m e I m pu l s e R e s pon s e Exchange to Exchange
Note: As for Figure 2. 25igure 5: Static Impulse Responses (Old Shares) . . . Exchange to Old Shares (Dec 1931)
Lags I m pu l s e R e s pon s e . . . . Exchange to GBPI (Dec 1931)
Lags I m pu l s e R e s pon s e . . . Exchange to Old Shares (Jun 1932)
Lags I m pu l s e R e s pon s e . . . . Exchange to GBPI (Jun 1932)
Lags I m pu l s e R e s pon s e Old Shares to GBPI (Dec 1941)
Lags I m pu l s e R e s pon s e . . . Old Shares to Exchange (Dec 1941)
Lags I m pu l s e R e s pon s ee
Lags I m pu l s e R e s pon s ee Note: R version 4.0.2 was used to compute the estimates.26igure 6: Static Impulse Responses (New Shares) . . . Exchange to New Shares (Dec 1931)
Lags I m pu l s e R e s pon s e . . . . Exchange to GBPI (Dec 1931)
Lags I m pu l s e R e s pon s e . . . Exchange to New Shares (Jun 1932)
Lags I m pu l s e R e s pon s e . . . . Exchange to GBPI (Jun 1932)
Lags I m pu l s e R e s pon s e New Shares to GBPI (Dec 1941)
Lags I m pu l s e R e s pon s e . . . New Shares to Exchange (Dec 1941)
Lags I m pu l s e R e s pon s ee
Lags I m pu l s e R e s pon s ee Note: R version 4.0.2 was used to compute the estimates.27igure 7: Static Impulse Responses (EQPI) . . . Exchange to EQPI (Dec 1931)
Lags I m pu l s e R e s pon s e . . . . Exchange to GBPI (Dec 1931)
Lags I m pu l s e R e s pon s e . . . Exchange to EQPI (Jun 1932)
Lags I m pu l s e R e s pon s e . . . . Exchange to GBPI (Jun 1932)
Lags I m pu l s e R e s pon s e EQPI to GBPI (Dec 1941)
Lags I m pu l s e R e s pon s e . . . EQPI to Exchange (Dec 1941)
Lags I m pu l s e R e s pon s ee