The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FOREIGN INVESTMENT FLUCTUATIONS AND EMERGING MARKET STOCK RETURNS: THE CASE OF MEXICO by John Clark and Elizabeth Berko Federal Reserve Bank of New York Research Paper No. 9635 November 1996 This paper is being circulated for purposes of discussion and comment only. The contents should be regarded as preliminary and not for citation or quotation without permission of the author. The views expressed are those of the author and do not necessarily reflect those of the Federal Reserve Bank ofNew York of the Federal Reserve System. Single copies are available on request to: Public Information Department Federal Reserve Bank of New York New York, NY 10045 FOREIGN INVESTMENT FLUCTUATIONS AND EMERGING MARKET STOCK RETURNS: THE CASE OF MEXICO John Clark and Elizabeth Berko• Federal Reserve Bank of New York 33 Liberty Street, Rm 737 New York, NY 10045-0001 (212) 720-6840 (212) 720-6246 (Fax) john.j.clark@frbny.sprint.com First Draft: November 1995 This Draft: November, 1996 • Mr. Clark is a senior economist in the Emerging Markets and International Affairs Group of the Federal Reserve Bank of New York and Ms. Berko was an assistant economist in the Emerging Markets and International Affairs Group of the Federal Reserve Bank of New York at the time this paper was prepared. The views and opinions expressed in this paper are those of the authors and do not necessarily represent those of the Federal Reserve Bank of New York or the Federal Reserve System. This paper should be regarded as preliminary and is not for circulation, citation, or quotation without the permission of the authors. We wish to thank Stijn Claessens, Gabriel de Kock, Kent Hargis and John Mullin as well as our colleagues at the Federal Reserve Bank of New York, including Richard Cantor, Arturo Estrella, David Laster, Frank Packer, and Marilyn Skiles for helpful comments and discussions. FOREIGN INVESTMENT FLUCTUATIONS AND EMERGING MARKET STOCK RETURNS: THE CASE OF MEXICO by John Clark and Elizabeth Berko• Federal Reserve Bank of New York November, 1996 Abstract We investigate the economically and statistically significant positive correlation between monthly foreign purchases of Mexican stocks and Mexican stock returns. We find that a I percent of market capitalization surprise foreign inflow is associated with a 13 percent increase in Mexican stock prices. We explore whether this correlation might be explained by permanent reductions in conditional expected returns resulting from expansion of the investor base along the lines modeled by Merton (1987), or correlations with other factors causing returns, price pressures, or positive feedback strategies by foreign investors, and conclude that the available evidence is consistent with the base-broadening hypothesis. • The views and opinions expressed in this paper are those of the authors and do not necessarily represent-those of the Federal Reserve Bank of New.York-or the·Federal Reserve. System. This paper should be regarded as preliminary and is not for circulation, citation, or quotation without the permission of the authors. FOREIGN INVESTMENT FLUCTUATIONS AND EMERGING MARKET STOCK RETURNS: THE CASE OF MEXICO' The current decade has witnessed a remarkable turnaround in investors' attitudes toward foreign stocks. As discussed by French and Poterba (1991), Lewis (1994), and Tesar.and Werner (1995), at the start of this decade investor portfolios demonstrated strikingly, indeed puzzlingly high weightings toward home country equities. Yet, in the 1990s investors have apparently awakened to the benefits of greater international diversification. For example, from 1989 through 1995 U.S. investors purchased foreign shares at 25 times the rate observed during the previous ten years. The change in flows has been particularly dramatic with respect to the socalled '!emerging markets" of Asia and Latin America, where foreigners have moved in recent years from holding almost no shares to a sizeable proportion of the market The literature on equity market segmentation (see Stulz (1995)) implicitly suggests that such dramatic changes in the investor base for emerging market equities should have profound · implications for their pricing: in particular, because of greater risk sharing and increased liquidity, expected returns should fall and prices should rise. In this paper we test this "base broadening" hypothesis--that foreign inflows cause emerging equity prices to rise--using monthly data from Mexico, the emerging economy that has received the largest amount of such inflows during the last seven years. Our discussion below is • The views and opinions expressed in this paper are those of the authors and do not necessarily represent those of the Federal Reserve Bank of New York or the Federal Reserve System. This paper should be regarded as preliminary and is not for circulation, citation, or quotation without the permission of the authors. organized as follows. We begin by briefly discussing ilie theoretical basis for the base-broadening ! hypothesis, placing particular emphasis on the model of barket segmentation proposed by Merton I (1987). We then lay out our econometric strategy for q~antifying the size of the base-broadening effect. For the most part, we apply the methodology us~d by Warther (1995) to analyze the • I • impact of mutual fund flows on U.S. stock and bond pri1es. We extend Warther' s approach, however, by broadening his set of tests for distinguishin~ the base-broadening hypothesis against several relevant alternatives that might account for the ptsitive correlation, evident in the data, between inflows and returns. The next two sections dis iuss the sources of data used for the study 1 and briefly overview the pattern of equity inflows to Mefco; further details are provided in the appendix. We then discuss the results of our econometri~ tests. We present our summary and conclusions in the final section. I 11 I. The Base-broadening Hypothesis I The theoretical literature on market segmentation: suggests two impo~t reasons why an exogenous increase in foreign participation could have p °\pelled Mexican share prices to a. permanently higher level, other things being equal. Broa~ening the investor base increases I diversification and risk sharing, lowering the required ris~ premium for Mexico specific volatility (Merton (1987) and Errunza and Losq (1985 and 1989) ). I Additionally, the influx of new . I investors could have lowered the perceived liquidity risk ~f Mexican stocks (Pagano (1989), Allen I and Gale (1991), and Hargis (1995.)) 1 . I Merton (1987) provides an intuitive and tractable ~ode! for illustrating how broadening the investor base for a given stock, and by extension fort emerging equity market, may raise 2 equity prices through risk pooling. In Merton's model, investors are assumed to invest in only an exogenously determined subset of the universe of equities. Merton characterizes the assumed barriers that prevent investors from holding fully diversified portfolios as informational, i.e. investors invest only in stocks about which they are "informed." However, he notes that his approach is consistent with explaining the impact of other barriers such ~ institutional restrictions . including "... limitations on short sales, taxes, transactions costs, liquidity, and imperfect divisibility of securities. "2 Merton demonstrates that in his framework if investors were able to invest in (that is, were "informed" about) all equities, the standard Capital Asset Pricing Model pricing relations would prevail, that is, the expected return on a given share (or market) would be a function of its covariance with the global market, but not its variance. However, with segmentation restrictions, the expected return on a share with a restricted investor base will be higher than its unrestricted return by a risk premium that will be an increasing function of the stock's conditional variance, the narrowness of the investor base and investors' risk aversion. Stocks (and, by extension, markets) with narrow investor bases exhibit higher expected returns because for the holders of these shares the variance of the returns on the stocks is more systematic than it appears from the perspective of the market as a whole. Specifically Merton shows that (I) where E(R.) is the equilibrium expected return on the kth security in the segmented market case, E(R *.) is the expected return in the complete absence of segmentation restrictions, R is the rate of return on the risk free asset and J... is the shadow cost of the segmentation restriction. A• in tum is shown to be equal to: 3 (2) I where 6 is the coefficient of aggregate risk aversion, o.tis the idiosyncratic component of the i variance of the kth stock's return, x, is the kth stock's s~are of the aggregate market portfolio, i and q, is th~ share of the total investor universe that inv~sts in (and is "informed" about) stock k. Merton also derives the comparative static result that~ the investor base increases (i.e. q, rises), I equilibrium required returns fall, and prices (PJ rise; th t is,: 0. = (3) In the Mexican context, Merton's q would me~ure the ratio of the number of investors that are "informed" about Mexican stocks to the total n~mber of investors: I i q = nm +nf I, (4) N ! where nm is the number of Mexican investors (who are ~l assumed to be "informed" about Mexican stocks), n} is the number of foreign investors 'jinformed" about Mexican stocks, and N is the total number of investors. Though we lack direct orservations on q, we can infer changes in q from changes in foreigner's holdings of Mexican stoc~ as a percentage of Mexican market I capitalization, defined as 6, assuming the number of do estic investors is held constant. Merton's model implies that if domestic and "informed' foreign investors have the same 4 infonnation sets, they will allocate their portfolios equivalently. Hence, 6 will equal the ratio of "infonned" foreign investors to the total of "infonned" foreign investors and Mexican investors: e= (5) For example, if foreign investors account for one-half of the 'investors that are "informed" about Mexico, according to Merton's model they would be expected to account for one-half of the holdings of Mexican stocks. By substitution and rearrangement, q can be rewritten, as follows: (6) This expression implies that: aq ae q 1-6 (7) Substituting into equation (3) from the main text, and rearranging, we have that: (8) Finally, noting that the foreign share, 6, will change with net foreign purchases of Mexican equities (Np.) as a percentage of Mexican market capitalization (MCap,_ 1), we can conjecture the following testable hypothesis based on Merton's comparative statics. In a regression of the fonn: 5 Return, = Po + f_P, P1 MCa (9) P,-1 ! ' ' where Return, is the return on Mexican stocks during m!onth t, we would expect to reject the null I . ' that the coefficient on net purchases, P,, equals zero in ~avor of the one sided alternative: P, > 0. 3 The specification of equation (9) parallels the basic fo~ of the regressions in W arther' s (1995) study of mutual fund flows and U.S. stock and bond ret~s.4 ' ' A. Expected versus unexpected inflows Even if the base-broadening hypothesis is true, te P, coefficient from regression (9) would likely understate the impact of foreign inflows onl Mexican equity prices. The efficient markets hypothesis implies that relevant information aviµIable at the start of the period should ! ' already be reflected in the price of assets at the start of te period. Hence, if foreign demand is expected to ultimately push prices to a higher equilibriu level, but foreigners only invest gradually, prices should rise ahead of the actual inflows Moreover, if investors are unsure of the magnitude of new foreign demand for Mexican stocks, e arrival of new information that causes investors to raise their estimate of total foreign inflows +ould push prices to a higher level. 5 ' These considerations suggest that instead of reg*ssing returns on actual changes in the I' investor base, we should regress returns on expectation* revisions about the evolution of the I ' . investor base. Unfortunately such expectational revisio~s are not directly observable. However, I if we assume that investors forecast the future evolutio1 of the investor base by studying realized flows, i.e. if: 6 (10) where st,+1 are foreign holdings of market kin period t+I, 7( ) is the forecasting function, the ai and bj are coefficients from the distributed lag processes A(L) and B(L) from the time series for net purchases: Np, = A(L)Np,_1 + B(L)0,_ 1 + e, and n,_ 1 is (11) a vector of other variables useful for forecasting net purchases, then innovations in the time-series of net purchases can serve as a proxy measure for expectational revisions to the time path of investor inflows. Hence, if we define unexpected net purchases, Ui(NP/MCap,_ 1), as being equal to observed purchases minus investors' expectation of net purchases, E,(NP/MCap 01 ), i.e. NP, NP, MCap,_ 1 MCap 1_ 1 u [---] - - - 1 - E, [ NP I MCap 1_1 l (12) we would expect returns to co-vary with unexpected purchases, but not with expected purchases. That is to say, in a regression of the form Return, = (13) we would expect to find that returns would show significant co-variance with unexpected inflows, but less co-variance with expected inflows: P, > p2 ~ 0. Strictly speaking, we might expect P2 to be equal to zero if we were able to accurately measure investors' expectations. However, given our short sample size, we are limited to within sample estimation of the forecasting model for expected inflows. Hence, we may ascribe· to 7 -• investors an overly accurate understanding of the time ~eries process for net purchases. In that case, our estimated value of expected net purchases mar include some unexpected net purchases, and we would reject the hypothesis that p2 = 0. I B. Disti~guishing the base-broadening hypothesis fro,,J, relevant alternatives I , We consider a number of alternative hypotheses Ithat can be advanced to explain the correlation between net foreign purchases and Mexicanttock returns. We focus on these hypotheses because of their plausibility as well as the fa t that they have other testable implications that potentially allow us to discriminate beteen them and the base-broadening hypothesis. Below we discuss our strategy for distinguifhing .between the base-broadening hypothesis and, respectively, the price pressure hypothe!is, the omitted variables hypothesis, and ! 1he positive feedback hypothesis. Table I summarizes ~e testable implications of these alternative hypotheses which we develop in more detail below. C. Alternative I: The price pressure hypothesis j Considerations based on improved risk sharing *d increased liquidity suggest that increased foreign participation should produce, ceteris pfmbus, a permanent reduction in risk . i premia and, hence, a permanent price rise. An alternatiye theory, which W arther (1995) refers to i as the price pressure hypothesis, suggests that the rises i~ prices associated with inflow surges are due to temporary illiquidity; such a theory would predict that inflow induced price increases would be subsequently reversed. For example, Shleiferl(l986) and Harris and Gurel (1986) ! presented evidence that increases in stock prices resulti~g from the announcement of inclusion of j individual stocks in the S&P 500 index are at least parti~ly reversed over the subsequent 30 to 60 I I 8 trading days. 6 The hypothesis of temporary illiquidity would appear plausible in the context of an emerging market such as Mexico. To test for the existence of such reversals, we adapt the approach of Warther and add lagged values of surprise inflows for the previous three months to our regression equation: 7 U NP,.1-; 1+1-lMC .) ap,_; (14) If the price-pressure hypothesis is true, we would expect to find that lagged surprise inflows have significant negative coefficients. In particular, we would expect to reject the null hypothesis (15) In the limit, the sum of the coefficients might equal the positive coefficient.on contemporaneous inflows. D. Alternative 2: The omitted-variables hypothesis If inflows are correlated with other factors that also move prices, then the correlation between inflows and prices may reflect primarily the influence of these third factors, and not the independent effect of foreign portfolio shifts. To control for possible omitted variable bias, we added several additional regressors, discussed in Section II.A, that help explain Mexican equity returns: Return,= Po + P, U,(NP /MCap,. 1) + PZ where z is a vector of additional regressors and (16) pis a vector of coefficients. Under the joint hypotheses that the omitted variable hypothesis is true, and that we have correctly included the 9 variables for which inflows were acting as a proxy, we y,ould not expect to reject the null that the ! I coefficient on surprise returns in a multi variate regressipn is equal to zero: 1 (17) Alternatively, if the base-broadening hypothesis is true, ~e would still expect to be able to reject I this null hypothesis, although the magnitude of the coef,cient on inflows, I p1, might change. . . While this approach of adding regressors potenti~ly might allow us to dismiss the correlation between inflows and returns as being driven by third factors, it must be admitted that it cannot provide definitive proof of the converse. The pofsibility exists that some other omitted I variables could be playing the hypothesized role of drivjng both inflows and returns. The power ' I of this test is clearly a function of the proportion of the yariation in returns that we are able to ! explain through the addition of these extra explanatory yariables. I I E. Alternative 3: The positive feedback hypothesis i If foreign investors are reacting to recent movenjents in Mexican prices by buying on increases and selling on declines (i.e. following a positi~e feedback strategy), then, because of 1 temporal aggregation, we might find a statistically significant correlation between inflows and . I. contemporaneous returns even if inflows are not causin~ returns. As noted by Warther, an I approach that could potentially uncover such feedback (rading would be to regress surprise i inflows on lagged returns, to see whether lagged return~ appear to have significant explanatory power. However, the power of this test might not be 1fited if investors' feedback horizon is quite short compared to the frequency of the available ~ata, for example a feedback horizon of I' I one or two weeks, when one is constrained to using onlr monthly data. ' 10 i ! Wart.her confronted the same problem in his analysis of U.S. mutual fund flows and proposed a solution, which we follow here, that takes advantage of the fact that return data are available at a higher frequency. Wart.her noted that if the positive feedback hypothesis is true, then there should be a higher correlation between flows over the whole month and returns during the weeks at the beginning of the same month and end of the previous month, than between monthly flows and returns during the last week of the same month. Hence, Wart.her constructed series for returns during the first to fourth week of each month and regressed unexpected inflows on returns during each of the four weeks of the same and the preceding months. Similarly, we tested for feedback-effects by regressing surprise inflows on the sum of the returns over each of the four weeks of the same month, as well as on the returns for each of the weeks 1 to 3 of the current month, and weeks 3 and 4 of the previous month: U,(NP /MCap,_ 1) = Po+ P1Sum, + where P2W3, + P3W2, + P4Wl, + P5W4,_ 1 + P6W3,_ 1 Sum,= W4,+ W3,+ W2,+ WI, (18) (19) and we follow Wart.her in defining week 1 returns (WlJ as the returns over the first seven calendar days of the month, week 2 returns (W2J as the returns over the second seven calendar days, week 4 returns (W 4J as the returns over the last seven calendar days, and week 3 returns (W3J as the return over the seven calendar days ending with the beginning of week 4. 8 Under the positive feedback hypothesis, we would expect to reject the null of zero coefficients on the individual weekly returns: (20) In contrast, the base broadening hypothesis--which would predict an equal impact for each of the weeks in the current month and no impact from last month--would not lead us to expect rejection: 11 II. Data sources and definitions Table II provides a summary description of the qata series used for this study and includes I the respective first and second sample moments. We m~asure foreign investment in Mexican shares from monthly flow data on net purchases of Mexjcan equities by foreigners; the data are I collected by the Central Bank of Mexico and published iy the National Banking and Securities I • Commission. Our sample, from January, 1989 to the Mfch, 1996, covers the entire period during which Mexico experienced significant equity inflows. 4ppendix I discusses this data in further detail and compares it with alternative measures of forei~n investment in Mexican equities. I For certain analyses, we separate the Mexican fltw data into publicized and unpublicized flows. Publicized inflows were identified from accounts! of specific transactions published in the ! financial press (primarily International Financing Revie~, see Table ill); such flows account for i i about two-fifths of the cumulative foreign equity purch¥es over our sample period, but only onefourth since mid-1992 (Table IV). Unpublicized inflowJ are defined as the difference between i total inflows and publicized inflows. The market capitalization statistics and stock pripe index used for this study are collected and published by the Mexican stock exchange. Mexicai) stock returns are measured using the i • percent change in the end-of-month IPC index in nominitl pesos. 9 The IPC index is a I capitalization weighted index of the prices of the largestl and most liquid Mexican stocks. A. Additional variables I Our additional regressors for the omitted variabl~s test comprised measures of the I following: movements in foreign (non-Mexican) stockprcesi the level and change in Mexican .• 12 short-term interest rates, the percentage change in the peso/dollar exchange rate, the within-themonth volatility of Mexican stock prices, shifting assessments of Mexican country risk, and a measure of revisions to aggregate earnings forecasts for Mexican stocks. We also experimented with measures of changes in foreign interest rates and/or bond yields but found that the explanatory power of these variables was weak and in any event subsumed by the included variables. We included two measures of changes in foreign stock prices (the percent changes in the Morgan Stanley Capital International (MSCI) World stock price index and in the S&P 500 price index) because we conjectlH'ed that although Mexican stock prices should respond to global price swings, they might be more closely linked with the U.S. market. The level of Mexican interest rates on one-month government bonds at the start of the period is included to capture the market's ex-ante expected nominal return in pesos. 10 The inclusion of the contemporaneous change in Mexican interest rates and the peso/dollar exchange rate controls for changes in the outlook for Mexican monetary and/or exchange rate policy during the month. The Merton model predicts that Mexican equity prices should co-vary negatively with their own variance, hence we used daily stock price data to construct a within-the-month measure of annualized Me~ican stock price volatility. Our country risk measure, the adjusted percentage change in the average "stripped yield" on Mexican long-term (Brady) bonds, was included as a broad gauge measure of shifts in investors' assessment of Mexico's access to spontaneous capital flows, and the country's overall growth prospects. 11 The stripped yield measures the yield to maturity on the uncollateralized portion of long-term Mexican bonds, and varies over time with both the level of global interest 13 rates and changes in spreads. 12 To separate the effects ~f changes in spreads from changes in global interest rate conditions, we calculate the "adjust1I'' percent decrease in the stripped yield (y,), defined as follows: (21) where y,-is the stripped yield at time t, and r, is the yield! on comparable maturiiy U.S. long-term treasury bonds. Measures of the revision to analysts' earnings fdrecasts were constructed from average i price-earnings (pier) forecasts for Mexican equities com~iled by 1/B/E/S for the period mid-1992 I to March 1996. The forecasts are available on a fiscal iear basis; to approximate a constant 12 month forward looking horizon we constructed a shifti~g weighted average of p/e forecasts for ' the current and next fiscal year. The earnings revision (~arnRev,) is obtained by multiplying the ratio of p/e forecasts by the ratio of price indices, i.e. I EarnRev, = [(pli!),. 1 I (pie'), "' (e1, I ef,_1) - I f. [p, I p,_ 1) - I I (22) ' I We proxy for aggregate earnings forecasts revisions dufng earlier periods (1989 though mid' . 1992) by using the conventional p/e ratio, that is, price+ trailing earnings, in place of pier in the above expression. I Time series on end-of-month U.S. interest ratesjbond yields and stock returns, and Mexican domestic currency government bill (Cetes) an exchange rates were taken from DRI. The Morgan Stanley stock price index was obtained fro/n Bloomberg News Services. The stripped spread data are from J.P. Morgan Securities. 14 III. Foreign Net Investment in Mexican Stocks: An Overview Significant foreign investment in the Mexican stock market basically dates from 1989 and coincides with major changes in the climate for foreign investment (Table IV). Earlier, foreign investment in Mexican equities appears to have been restrained by investors' general reluctance to diversify internationally, and their concerns about restrictions, actual and potential, on foreign . . equity investments in Mexico. This situation was changed, on the one hand, by developments in the U.S. and other industrial countries--institutional changes and low bond and dividend yields-that encouraged investors to look outward, including to Mexico (Calvo, Leiderman and Reinhart (1993 and 1994), Chuhan, Claessens and Mamingi (1993), and Gooptu (1993)). On the other hand, the Mexican authorities' implementation of structural reforms--such as expanding the range of companies and shares in which foreigners could invest--and pursuit of macroeconomic policies improved both the attractiveness and feasibility of investment in Mexico (Loser and Kalter (1992)). Evidence of the Mexican government's success in restoring investor confidence through macroeconomic and structural policies can be seen in the sharp reduction in the risk premium on long-term Mexican bonds from end-1988 to end-1993 (Chart 1). 13 Concentrating on the Mexico-specific factors that may be less familiar to the reader, in 1989, the Mexican authorities modified the foreign investment regime to ease restrictions on foreign participation in the Mexican stock market and, in 1990, removed a 40 percent tax on dividend income (Mullin 1993). 14 Also, the government moved aggressively to divest itself of holdings in a number of key industries, chiefly banking and telecommunications, raising some $25 billion through the sale of state owned enterprises between 1989 and 1994. While only a fraction ofthese privatized shares were sold directly to foreign investors--Telmex, the telephone company, 15 standing as the major exception--foreigners often boughjt into the newly privatized firms through the secondary market, or during subsequent public offetjngs by the privatized firms. I A. Size and volatility offlows I I The scale and volatility of equity inflows to Me*co has been quite large. Between 1989 and end-1993 foreigners increased their share of Mexic:\n stocks from negligible levels to over I one-fourth; the share of foreign ownership has fluctuateµ around this level subsequently (Chart 1 and Table IV). Inflows into the Mexican stock market ~enerally have been positive, averaging 0.42 percent of market capitalization per month, but ha~e varied considerably in intensity from month-to-month with a standard deviation of 0.63 perc~nt of market capitalization, and maxima i and minima of 4.3 percent and -0.3 percent of market c*pitalization (Chart 2, Table II). Actual outflows have occurred only infrequently, generally dutjng periods of heightened policy I uncertainty. For example, foreigners sold in 1994 folloting the assassination of the presidential i front0 runner (April) and ahead of the December inaugu~ation of the new president. Notably--and contrary to some popular perceptions--the market backllish that followed Mexico's December ! I 1994 devaluation produced only limited realized outflmf,s from the equity market. We note that aggregate mutual fund inflows int~ U.S. stocks, which have garnered much attention of late in the financial press, averaged 0.052 p~rcent of U.S. stock market capitalization I during the 1984 through 1992 time period studied by \\farther (1995), with a standard deviation of only 0.078 percent of market capitalization. Hence, by the metric of market capitalization, I foreign flows into Mexico's stock market were eight titjies as large and eight times as volatile as I mutual fund flows into the U.S. stock market. 16 IV. Test Results A. A first pass: returns vs. total equity inflows As shown in Table V, there is a strong correlation between inflows into the Mexican equity market and contemporaneous price performance. As can be seen, foreign inflows explain about one-seventh of the monthly variation in peso returns over the whole seven year sample period, 1989-1996:03. Column (1) reports the results of regressions of stock market returns measured in local currency on contemporaneous total monthly foreign inflows as a percentage of market capitalization. Virtually identical results are found in Columns (2) and (3) when we used local currency excess returns (local currency returns minus the local government bill rate) or returns measured in dollars. Because we generally found equivalent results throughout whether we used local currency, foreign currency, or excess returns, below we tend to report on tests using local currency returns as the dependent variable so as to maintain symmetry with the specifications used by Warther. The regression coefficient--foreign purchases of 1 percent of market capitalization are associated with a 6 percent rise in the Bolsa index--is quite large when compared with some studies of the Price Pressure-Upward Sloping Supply Curve literature. These latter papers look at price responses for individual shares to shifts in demand; for example, Harris and Gurel (1986) and Shleifer (1986) found that following announcements that new companies would be added to the S&P 500 index, the price of their shares typically rose by around 3-4 percent, presumably in response to purchases by index fund managers who typically purchase around 3 percent of the shares of the included stocks. Similarly, Bagwell (1991) reported evidence from Dutch auctions .,, , thatsuggested·prices rise by 1.7 percent for each I-percent of.outstanding shares purchased. On,,,. 17 the other hand, Warther (1995) found a much larger response coefficient in his regressions of changes in U.S. stock prices on surprise inflows into eq*ity mutual funds; in his data a surprise ! inflow to stock mutual funds equal to 0.1 percent of ag!!fegate market capitalization was I associated with a price rise of 5.2 percent (Table 4, p. 2f3)--nearly ten times the price rise found in our data. Our coefficient is also considerably smallerlthan that found by Jun (1993) in his study of foreign inflows into the Korean equity market. Jun f~und that a $1 billion net foreign purchase ' ' of Korean equities, roughly I percent of market capitalifation was associated with a 24 percent price rise; however, his results must be viewed as tenta~~e given the short sample period (16 months). B. Anticipated vs. surprise inflows To test whether inflows respond differently to anticipated and surprise inflows, we first constructed a forecasting model for anticipated inflows, !adapting the approach used by W arther (1995) to analyze mutual fund flows. Specifically, we rJst separated inflows into the Mexican equity market into two components, publicized inflows $d unpublicized inflows, and then fit a 1 time series model to explain unpublicized inflows. We t~rmed the residuals and fitted values from the final model surprise unpublicized inflows and expecJii unpublicized inflows, respectively. We I treated publicized equity placements separately because ~ uch flows clearly do not represent ' surprise inflows in the months that they are concluded. typically these flows are reasonably well known to the market at least several months in advance, !although the exact timing is not always easy to identify. I In·his study of mutual fund flows, Warther used rnly lagged.inflows to forecast current 18 - inflows. In our case, we experimented with including current and lagged publicized inflows as well, to control for possible crowding out of unpublicized flows by publicized flows. To guard against over fitting our model, we then experimented with reduced lag lengths and dropping the publicized flows. As shown in Table VI, using the Schwartz criterion, the best results were obtained with a simple AR(!) specification, while the Akaike criterion favored an AR(2) specification. Using either criterion, current and lagged publicized inflows were not found to be useful in forecasting unpublicized flows and therefore were excluded. In the remainder of this paper, we used the AR(!) specification for separating inflows into the anticipated and surprise components. However, we found similar results for all the tests reported below when we used the AR(2) specification, or when we broadened the set of prediction variables as would be suggested by a more structural model. We found evidence that generally supported the conjecture of differential responses when we tested whether returns appear to respond to anticipated or anticipated flows. As shown in Table VII, Columns (I) and (2), when total equity inflows are replaced with surprised unpublicized inflows, the coefficient on inflows rises from 6 to 13, and the regression R 2 increases, a result consistent with the conjecture that the inclusion of expected flows biases our coefficient on total inflows toward zero. However, when we include all three component s of equity inflows as reported in Column (3), i.e. publicized inflows, expected unpublicized flows and surprise unpublicized flows, we find that our hypothesis that the coefficients on the first two · variables is equal to zero is rejected at the 5 percent level. Nonetheless, the estimated coefficient on the surprise inflows is higher than the estimated coefficients on the more predictable components, as would be predicted by the base-broadening/efficient markets hypotheses,.and we.... 19 are able to reject at the 1 percent level the hypothesis o( equal coefficients on the various components of the equity inflows. The significant posi~ve correlation found between publicized ' ' inflows and prices could reflect market timing by Mexi4an issuers, i.e. issues are not launched I I when prices are declining. We also find that the result <!>f non-zero coefficients on expected flows I • I is strongly influenced by the positive association betwe~n prices and inflows during the first I I . Telmex deal. When we include a dummy for the May 11991 Telmex placement, we are then unable to reject the hypothesis that the coefficients on ppb!icized inflows and expected I unpublicized inflows are equal to zero, and we still are ible to reject the hypothesis of equal coefficients on surprise and ,expected inflows at the I p~rcent level. i C. Are the price rises associated with inflows temporary or permanent? I Our test to distinguish whether the data suppo~ the Price Pressure Hypothesis as an alternative to the base broadening hypothesis, failed to 1etect evidence of price reversals. While I the hypothesis of price pressures emerging from tempor~ illiquidity would appear plausible in I I . the context of an emerging market such as Mexico, partjcularly in view of the large coefficient on surprise inflows, we did not find evidence that lagged s~rprise inflows were associated with I negative returns. On the contrary, the coefficients on la~ surprises summed to an economically significant but statistically insignificant positive numberl(Table VIII, Column (2)). In particular, i the previous month's surprise was found to carry an eco~omically significant positive coefficient. I Moreover, when we included only the previous month'~ inflows, we found that we could reject I I the null of a zero coefficient at the 15 percent level in faivor of the one-sided hypothesis of a i positive coefficient (Column (3)). We note that Warther also found weak evidence, (Table 5, p. ~• 20 226) that surprise inflows this month are associated with positive returns next month. We conjecture that this delayed reaction weakly evident in both data sets could be the result of learning by market participants about the true magnitude of the initial shift in investor demand. D. Do additional variables explain away the correlation between flows and returns? To control for possible omitted variable bias, we added several additional regressors, discussed in section II.A, that help explain Mexican equity returns. As can be seen in Table IX, our additional variables explain a significant share of Bolsa returns. As a group, the variables explain about five-ninths of.the monthly peso variation in Mexican stock returns (Column (1)) and two-thirds of the variation in dollar returns (Column (2)). All of the variables possess their expected signs and almost all of the variables are found to have statistically significant independent explanatory power. The adjusted change in stripped spreads--which captures shifts in market sentiment toward Mexico-specific risk--has the greatest explanatory power, by itself accounting for about one quarter ofthe variation in Bolsa returns. Controlling for these third factors reduces but does not eliminate the estimated impact of foreign flows on prices (Table X). As can be seen, the estimated coefficient on surprise foreign purchases falls by about two-fifths when we expand out regressor set to include the variables in Table IX, but remains significantly different from zero at the 1 percent level. Overall, the model of returns specified in Table X including surprise inflows explains more than three-fifths of the variance of local currency returns and about 70 percent of the variation in dollar returns. 21 E. Do foreign investors chase recent returns? We found no evidence that foreign investors are ~ositive feedback traders. As shown in ' column ( 1) of Table XI, lagged monthly returns do not ~elp forecast unpublicized flows. Likewise, when we used weekly returns, we did not find! any evidence that inflows are more I positively associated with returns during the weeks at thF beginning of the month or the end of the previous month, contrary to the predictions of the positiye feedback hypothesis, Instead, as shown in columns (2) and (3), our highest coefficient es~ate in a regression of surprise flows on weekly returns is found on the return on the fourth wee~ of the current month and the lowest ' I coefficient for the current month is found on the return from week 1. This pattern exactly i reverses of the predictions of the positive feedback hypo~esis. However, these difference within ' i the month are not statistically significant. As shown in 9olumn (3), we cannot reject the null of j equal coefficients on each of the four week's of the currtnt month, and zero coefficients on the returns from the last two weeks of the previous month. V. Summary and Conclusions The results from our study of the inflow-price litjkage support the belief, common among . '. market participants, that foreign inflows to emerging eq*ity markets have an important impact on ' emerging equity returns. Surprise foreign purchases totlj.lling one percent of market capitalization (a three standard deviations innovation) are associated 1ith contemporaneous price rises of about 13 percent. We considered a variety of hypotheses to accou* for the correlation between flows and returns in our data set and conclude that the evidence is !tonsistent with the base-broadening 22 hypothesis, that is, the hypothesis that greater risk sharing and improved liquidity resulting from foreign inflows produce permanent price rises. We did not find evidence of inefficiency in the response of prices to inflows: forecastable inflows do not produce statistically significant price effects, but surprise inflows do; and we did not find any evidence that inflow-associated price changes reverse themselves in subsequent months. We found that we could npt explain away the price-inflow link through the inclusion of other variables that also cause returns. Inclusion of additional regressors that explained five-ninths of the monthly variation in local currency returns and two-thirds of the variation in dollar stock returns reduced the coefficient on surprise inflows by about two-fifths, but the inflow coefficient remained highly significant. Finally, tests of whether price-inflow correlation is the result of foreign investors chasing recent price rises produced negative results. There are considerable similarities between our findings and those from W arther' s (1995) study of U.S. mutual fund flows and stock returns. Warther found that from 1984 to 1992 mutual fund flows explained a significant fraction of monthly U.S. equity returns. Similar to our . findings for Mexican returns, he found no evidence of return reversals or positive feedback trading. Compared with our study, Warther found a higher response coefficient (52 versus 13) on surprise inflows when scaled by market capitalization. However, mutual fund flows into U.S. stocks were smaller on average (one-eighth as large in relation to market capitalization) and less volatile than foreign inflows into Mexican equities. Also, W arther found that inflows into U.S. stocks explained a greater fraction of return variability, an R 2 of 1/2 vs 1/7 in our data set. This high explanatory power might reflect the role of omitted variables. W arther did not control for '81her factors-causing returns that could have been correlated with mutual.fund flows. 23 It is not clear whether the coefficients found in t1te present study will necessarily generalize to other cases, or even prove stable over tim4 for Mexico, nor are we able to provide definitive proof against the possibility that our estimate4 coefficients overstate the impact of demand shifts on prices due to correlation with omitted ~ariables, although we were able to eliminate some obvious candidates. The evidence sugg~sts, however, that analysts seeking to model emerging market returns may wish to take into acjcount fluctuations in foreign equity portfolio investment activity. Ignoring such flows may fause analysts to observe returns in recent I years that appear to be too high or too uncorrelated wi~ global equity returns; this behavior, which might reflect a disef}tlilibrium adjustment to increr5ing market integration, potentially could be viewed as symptomatic of a continued equilibrium o~ market segmentation. 24 References Allen, Franklin and Douglas Gale, Limited market participation and the volatility of asset prices, American Economic Review 84, 933-955 Bagwell, Laurie S., 1991, Shareholder heterogeneity: Evidence and implications, American Economic Review 81, 218-221. Bailey, Warrep and Y. Peter Chung, 1995, Exchange rate fluctuations, political risk, and stockretums: Some· evidence from an emerging market, Journal of Financial and Quantitative Analysis 30, 541-561 Bailey, Warren and Julapa Jagtiani, 1994, Foreign ownership restrictions and stock prices in the Thai capital market, Journal of Financial Economics 36, 57-87. Bohn, Henning and Linda L. Tesar, 1996, U.S. equity investment in foreign markets: Portfolio rebalancing or return chasing? American Economic Review 86, 77-81 Calvo, Guillermo, Leonardo Leidennan, and Carmen Reinhart, 1993, Capital inflows and real exchange rate appreciation in Latin America, IMF Staff Papers 40, 108-151. Calvo, Guillermo, Leonardo Leidennan, and Carmen Reinhart, 1994, Inflows of capital to developing countries in the 1990s: Causes and Effects, mimeo Chuhan, Punam, Stijn Claessens and Nlandu Mamingi, 1993, Equity and bond flows to Asia and Latin America: The role of global and country factors, World Bank Working Paper Claessens, Stijn and Moon-Whoan Rhee, 1994, The effect of barriers to equity investment in developing countries, in Jeffrey A. Frankel ed., The Internationalization of Equity Markets (University of Chicago Press, Chicago, 11.), 231-276. Clark, John, 1994a, Debt reduction and market reentry under the Brady plan, Federal Reserve Bank of New York Quanerly Review, 38-62 Clark, John, 1994b, The structure, growth and recent perfonnance of the Latin American bond market, Federal Reserve Bank of New York Research Paper No. 9416 25 De Long, J. Bradford, Andrei Shleifer, Lawrence Summers, w>d Robert Waldmann, 1990, Noise trader risk in financial markets, Journal of Political Economy 98, V03-738 Erb, Claude B., Campbell R. Harvey, and Tadas E. Viskanta, 1995, Country risk and global equity selection, Journal of Ponfolio Management, 74-82. Errunza, Vihang and Etienne Losq, 1985, httemational asset p~cing under mild segmentation: Theory and test, Journal of Finance 40, 105-124. Emmza, Vihang and Etienne Losq, 1989, Capital flow control:s, international asset pricing, and investors' welfare: A multi-country framework, Journal of Finahce 44, 1025-1037. Eun, Cheol S., S. Janakiramanan, 1986, A model of intematiol)al asset pricing with a constraint on the foreign equity ownership, Journal of Finance 41, 89~-914. : Fama, Eugene F. and Kenneth R. French, 1988, Pennanent an4 temporary components of stock prices, Journal of Political Economy 96, 246-273. French, Kenneth R. and James M. Poterba, 1991, htvestor divtjrsification and international equity markets, ' American Economic Review, 81, 222-226. Gooptu, Sudarshan, Portfolio investment flows to emerging mju-kets, in S. Claessens and S. Gooptu, eds. Ponfolio Investment in Developing Countries, Worl4 Bank Discussion Paper No. 228, 45-77. Hargis, Kent, 1995, The internationalization of emerging equitt markets: Domestic market development or retardation?, University of Illinois Working Paper. i Harris, Lawrence and Eitan Gurel, 1886, Price and volume eff~cts associated with changes in the s&p 500 list: New evidence for the existence of price pressures,1Journa1 of Finance 41, 815-829. ' i Harvey, Campbell R., 1995, Predictable risk and returns in emtrging markets, Review of Financial Studies 8, 773-816. Hietala, Pekka T., 1989, Asset pricing in partially segmented n)arkets: evidence from the Finnish market, Journal of Finance 44, 697-718. 26 Jun, Kwang, 1993, Effects of capital market liberalization in Korea: Empirical evidence and policy implications, in S. Claessens and S. Gooptu, eds. Ponfolio Investment in Developing Countries, World Bank Discussion Paper No. 228, 404-426. Kadlec, Gregory B. and John J. McConnell, 1994, The effect of market segmentation and illiquidity on asset prices: evidence from exchange listings, Journal of Finance 49, 611-636. Lewis, Karen K., 1994, Puzzles in international financial markets, Universi~y of Pennsylvania Working Paper. Loser, Claudio and Elliot Kalter, eds., Mexico: The strategy to achieve sustained economic growth, IMF Occasional Paper No. 99, September 1992. Merton, Robert C., 1987, A simple model of capital market equilibrium with incomplete information, Journal of Finance 42, 483-510. Mullin, John, 1993, Emerging equity markets in the global economy, Federal Reserve Bank ofNew York Quanerly Review, 54-83. Pagano, Marco, 1989, Endogenous market thinness and stock price volatility, Review of Economic Studies 56, 269-287. Shleifer, Andrei, 1986, Do demand curves for stocks slope down?, Journal of Finance 41, 579-590. Stulz, Rene M., 1995, International portfolio choice and asset pricing: An integrative survey, in Jarrow, et al. eds. Handbooks in Operations Research and Management Science: Finance (North-Holland) 201223. Stulz, Rene M. and Walter Wasserfullen, 1995, Foreign equity investment restrictions, capital flight, and shareholder wealth maximization: Theory and evidence, Review of Financial Studies 8, 1019-1057 Tesar, Linda L. and Ingrid M. Werner, 1993, U.S. equity investment in emerging stock markets, in S.Claessens and S. Gooptu, eds. Ponfolio Investment in Developing Countries, World Bank Discussion Paper No. 228, 200-220 27 Tesar, Linda L. and Ingrid M. Werner, 1994, International equity transactions and U.S. portfolio choice, in Jeffrey A. Frankel ed., The Internationalization of Equity Markets (University of Chicago Press, Chicago, II.), 185-230. Tesar, Linda L. and Ingrid M. Werner, 1995, Home bias and high turnover, Journal of International Money and Finance 14, 467-92 Warther, Vincent A., 1995, Aggregate mutual fund flows and security returns, Journal of Financial Economics 39, 209-235. 28 Notes 1. If a stock market is characterized by low turnover and/or insider trading, investors may find it difficult to adjust their portfolios without moving prices against themselves; moreover the situation may be self-reinforcing, as some investors may react to such illiquidity by shying away from the market, further reducing liquidity. In such an environment, investors discount future earnings by an additional factor that takes into account the fact it may be difficult tp fully realize the fundamental value of a share (see Pagano (1989) Allen and Gale (1991) and Hargis (1995)). However, broadening the universe of active investors can increase the elasticity of demand at given prices, and lower concerns about liquidity risk. In fact, a number of Mexican stocks have become much more actively traded as a result of the internationalization of Mexico's equity market. In particular, Telmex, whose ADRs have traded on the New York stock exchange since 1991, consistently ranked among the most actively traded shares on the New York stock exchange in recent years; the high liquidity of Telmex has also improved the liquidity of other Mexican shares because investors have been able to use Telmex as a proxy for trading Mexican market risk. 2. In some countries, the government establishes a ceiling on the proportion or type of shares that foreigners may hold. In such a situation, modeled by Eun and Janakiramanan (1986), two prices may be observed, a price for foreign investors and a lower price for domestic investors. Such price spreads have been studied by Hietala (1989), Baley and Jagtiani (1994) and Stulz and Wasserfullen (1995) for Finnish, Thai, and Swiss stocks, respectively. 3. Our approach may be viewed as a time-series alternative to the event study methodology of Kadlec and McConnell, 1994. Kadlec and McConnell studied market reactions to announcements 29 of new listings on the New York Stock Exchange, and found that the price impact was related to the size of the associated increase in the investor base. 4. P, in equation (9) essentially measures the mean of the variables in parentheses in front of the ae term in equation (8). Some of the bracketed terms, in particular xk, the ratio of Mexico's market capitalization to global wealth, may vary over our sample period. We adopted the fixed coefficient specification reported'in this paper in part to maintain parallelism with Warther's (1995) functional forms, and also because of measurement error problems with respect to the correct measure of global wealth to put in the denominator. We did experiment with using the ·Morgan Stanley world stock,price index to construct a measure of xk. We found similar results whether we adjusted for changes in xk or assumed xk to be constant. 5. Merton recognized this point, and noted that the issue would have to be addressed in constructing a dynamic version of his model. Specifically he points out (p.500) that "if a favorable story implies an upward revision in those anticipations (i.e. expectations about the future time path of the size of the investor base), then the price should rise immediately, even if there is a time lag before the newly-informed investors take positions. Similarly, an unfavorable story implying a reduction in the anticipated growth in the investor base should cause an immediate price decline." The noise trader model of DeLong Shleifer, Summers and Waldmann (1990) gives an example where a noise parameter causes the investor base to shift randomly and where investors (and hence prices) take into account the mean and variance of such shifts in the investor base. 6. Fama and French (1988), on the other hand, have presented evidence of price reversals over a five-year period. Our sample is too short to detect reversals over such long horizons. 30 7. In contrast, Warther (1995) tested for the existence of a negative relation between flows and future returns by regressing monthly flows on the returns from the each of the four weeks of the current, previous and future month. Our approach is more general in allowing for reversals over a horizon as long as three months. 8. This approach results in the truncation of up to three days in the middle of the month, but accounting for these missing days does not materially affect the results. 9. Using annual data, the percent change in IPC price index has a 99.5 percent correlation with the International Finance Corporation's total return index for Mexico measured in local currency. 10. Harvey (1995) presents.evidence from a broad range of emerging markets that local interest rates help predict returns. 11. The importance of controlling for country risk has been suggested by Bailey and Chung (1995), who provide evidence of time varying risk premia for country risk, as proxied by movements in sovereign debt prices, from panel data of returns on individual Mexican equities. Also, Erb, Harvey and Viskanta (1995) show that much of the excess returns displayed by emerging equity markets is correlated with measures of country risk, i.e. high default risk countries (as measured from surveys of internationally active banks) exhibit higher expected returns. Erb, Harvey and Viskanta's results suggest that improving policy performance should have been associated with an increase equity prices. 12. Brady bonds are highly liquid partially collateralized bonds that were originally issued to commercial banks as part of the country's commercial bank debt reduction agreement concluded under the auspices of the Brady plan in 1990. Trading of Mexico's Brady bonds is centered in New York and takes place among a range of sophisticated investors that includes trading desks at 31 the largest U.S. and European commercial banks and securities firms, hedge funds and mutual funds, and financial institutions and high net worth investors from Latin America. For the period prior to the 1990 issuance of the Brady bonds we use changes in the secondary market price of Mexican bank debt. For further discussion of the liquidity and participants in the Brady bond market, see Clark I 994b. 13. The improvement in Mexico's market access was accelerated by the 1990 debt reduction agreement with the country's commercial bank creditors. For a discussion of the restoration of market access for Mexico and other restructuring countries, see Clark, 1994a. 14. After 1989, the most significant remaining restrictions applied to bank shares where foreigners were subject to a 30 percent ceiling on aggregate holdings. Claessens and Rhee (1993) estimate that the share of Mexican stocks that could potentially be held by foreigners jumped from 10 percent of the market in early 1989 to around 60 percent in 1990 and over 80 percent by 1993. 32 Table II. Description of Time Series Description Total net foreign purchases of Mexican equities Publicized net foreign purchases Unpublicized net foreign purchases Source Unite Mun Median Stindard deviation Banco de Mexico Percent of market capitalization Millions of dollars 0.42% 343 0.29% 216 0.63% 578 IFA, Bank of New York Percen1 of market capitalization authors' esti'mates Millions of dollars 0.13% 128 0.00% 0 0.49% 323 Total - Publicized Percent of market capitalization MIiiions of dollars 029% 215 023% 176 .0.35% 401 Expected unpublicized net foreign purchases See text Percen1 of market capital~atlon 0.30% 027% 0.20% Surprise unpubliclzed net foreign purchases See text Percent of market capitalization •0,00% 0.00% 0.30% Peso return on Balsa index Balsa de Valores Percent change 3.61% 3.97% 9.49% Dollar return on Balsa index Balsa de Valores Percent change 2.39% 3.44% 10.97% Excess peso return on Balsa Index Balsa de Valores Percent change 1.29% 2.08% 9.30% Percentage return on MSCI World dollar index DAl•McGraw/Hill Percent change 0.41% 0.79% 4.41% Percen1age return on S&P 500 DRl•McGrawlHIII Percent change 0.96% 1.17% 3.36% ShorMerm peso interest rate Banco de Mexico End-of-month yield, annualized 27.70% 20.49% 15.30% Change in shorMenn peso interest rate Banco de Mexico Change In yield ·0.14% •0,48% 5.43% Adjusted percent decrease in stripped yield of Mexican Brady bonds See text Percent change 0.93% 0.73% 7.62% Percent growth in aggregate earnings forecasts See text Percent change 2.32% 0.44% 11.03% Annualized daily stock price volatility See text Percent, period average 22.27% 19.71% 9.35% Banco de Mexico Percent change 1.63% 0.41% 5.91% Percent change in peso-dollar exchange rate Note: Sample comprises monthly data from January 1989 to March 1996. Variables scaled by market capitalization are scaled by market capitalization at the start of the period. Percent changes are all calculated on an end of period basis. Table Ill Mexico: Publlclzed International Equity Flows Millions of dollars Date Company Apnl 1991 Fomento Eonomico Mexicano Vitro S.A. Internacional de Ceramica Amount 87.4 37,0 22.7 Date Company August1993 Grupo Situr September 1993 Panamencan Beverage Co. Grupo Tnbaaa Coca-Cola Femaa 264.0 211.0 151.0 Amount 9,1 May 1991 Telmex July 1991 Grupo Gigante 48.8 October 1993 Grupo lndustnal Maseca 49,5 September 1991 Grupo Carso Empaques Ponderoaa 213.7 32.7 November 1993 Bufete lndustnal S.A. 95.8. October 1991 December 1993 Tubos de Acero de Mexico 41.0 November 1991 Vitro S.A. Aerovias de Mexico Grupo Video Visa Tranportacion Maritima Empaques Ponderoaa 165.0 95.4 45.0 35.0 33.0 Grupo Televiaa GF Serfin Grupo Mexicana de Desarollo Grupo Casa Autrey 874.8 308.3 248.5 63.8 January 1994 Grupo Tribasa 39.9 February 1994 Grupo Televisa Grupo Situ, Grupo Tribasa Empresas I.a Modema 747.0 51.0 300.3 2n.s March 1994 GF GBM Atlantico 90.4 GF Bancomer Sears de Mexico Grupo Posadas 602.0 102.0 28.0 April 1994 Grupo Embotellador de Mexico 119.4 June 1994 CemexS.A. Empresas !CA Grupo luaacell S.A. BanpaisS.A. 155.7 102.7 461.0 326.0 July 1994 May 1992 Telmex 1243.3 Grupo Industrial Durango Grupo Sidek DESCS.A. 111.3 96.7 55.2 June 1992 Tranportacion Maritima El Puerto de Liverpool 76.0 48.0 August1994 Corporacion GEO Grupo Mexicans de Desarollo 44.3 29.6 July 1992 Grupo Video Visa 20.6 September 1994 Sigma Alimentos 131.0 December 1992 Grupo Embotellador de Mexico 135.3 October 1994 HylsamexS.A. 123.5 February 1993 Grupo Carso 235.0 December 1994 March-1993 Grupo Simec Internacional de Ceramics Consorcio G Grupo Dina S.A. 173.2 49.3 22.7 February 1996 June 1993 Grupo Simec S.A. . Corporacion Industrial San Luis 65.7 45.0 March 1996 July 1993 Grupo Radio Centro Servicios Financieros Quadrum 45.6 40.6 Elamex Panamerican Beverage 28.4 44.4 December 1991 March 1992 April 1992 1876.2 Sources: International Financing Review, Bank of New Yori<, Bolsa Mexicana de Valores, authors' estimates Note: Excludes domestically placed tranches. Appendix 1: Measuring Foreign Investment in Mexican Equities Foreigners invest in Mexican shares through a variety of modalities, direct and indirect. Following the passage of the new foreign investment law in 1989, most classes of Mexican shares could be purchased and held by foreigners directly and without overall ceilings on foreign holdings. Class A Mexican shares cannot be held directly by foreigners; however these shares may be held indirectly through "ordinary participation certificates" (CPOs) issued by NAFINSA, the state-owned development bank, against A shares held in trust. Shares held to back CPOs are referred to as "Neutral Fund" holdings. CPOs entitle foreign investors to the cash flow associated with the shares but do not confer voting rights. Many individual and institutional foreign investors nonetheless have preferred to hold Mexican shares indirectly, via the purchase of American or Global Depository Receipts (AD Rs or GDRs). Foreign investors find ADRs attractive because they trade and settle outside of Mexico--most often in the case of Mexican ADRs on the New York Stock Exchange--and allow foreigners to bypass the Mexican foreign exchange market. Finally, many retail investors rely on an additional layer of intermediation, by holding their Mexican shares indirectly through mutual funds; these mutual funds in turn hold ADRs, free subscription or neutral fund shares. To measure foreign investment activity in the Mexican equity market we draw primarily on monthly flow data on net purchases of Mexican equities by foreigners collected by the Central Bank of Mexico. This data is published by the National Banking and Securities Commission on a monthly basis and is included in Mexico's official quarterly balance of payments estimates. These monthly Mexican flow data are available from 1989 to the present; the coverage since 1991 includes a disaggregation between net flows into ADRs, direct or "free subscription" holdings of 33 unrestricted shares, Neutral Fund holdings of restricted shares, and purchases by the Mexico Fund, a closed-end mutual fund. For certain analyses, we separate the Mexican flow data into publicized and unpublicized flows. Publicized inflows were identified from accounts of specific transactions published in the financial press (primarily International Financing Review, see Table ill); such flows account for about two-fifths of the cumulative foreign equity purchases over our sample period, but only onefourth since mid-1992. Two alternative data sources are available for tracking foreign participation in the Mexican stock market. Data on the stock of foreign holdings of Mexican shares are collected by the Mexican stock exchange (Bolsa de Valores) and are available on a monthly basis from end-1990. Holdings data disaggregated into four broad categories (the same categories as for inflows) are regularly reported in the financial press; the Bolsa also publishes more disaggregated data on foreign holdings on a stock-by-stock, and modality-by-modality basis. Flow data on U.S. net purchases of foreign equities are published by the U.S. Treasury Department. The U.S. data, which is collected on a gross and net basis, is collected monthly, but flows by country are only published quarterly; a monthly breakout of flows to Mexico was provided to us by the Treasury Department. On average, identified purchases of Mexican equities by U.S. investors account for four-ninths of the foreign equity inflows recorded by the Mexican authorities; discussions with market participants suggest an even larger role for U.S. investors, implying some degree of undercounting in the U.S. data. Chuhan, Claessens and Mamingi (1993), Tesar and Werner (1993 and 1994), and Bohn and Tesar (1996) have used the U.S. data to analyze the pattern U.S. portfolio flows to a range of countries including Mexico. 34 These three alternative sources for data on foreign investment in Mexican equitie s typically reveal similar trends and movements. However, the implicit flows derivable from fluctuations in the value of foreign holdings are often more volatile than the directly measured flows and have a correlation of only 55 percent with measured flows. The higher volatility of the derived flows likely reflects short-run divergence in the price performance of the foreign portfolio vis-a-vis the index portfolio, rather than a better measure of purchasing activit y. Overall, there is a reasonably close correlation between the Mexican and U.S. flow data (76 percent correlation), particularly since mid-1992 (89 percent); the U.S. flow data and the implied flows from the stock data correlate least well (35 •percent correlation). 35 Chart 1 Mexico: Selected Financial Time Series Foreign holdings of Mexican Equities Mexican Local Currency Interest Rates Percent of Market Gapltallzation Annualized yield In percent :a} BO 25 60 ---------------------------- 20 15 10 oi..._ _.__....,_ _.,.__....._ _.....__ _ _ _ _........, 1989:01 1990:01 1991:01 1992:01 1993:01 1994:01 1995:01 1996:01 r-01 199<>01 1991:01 1992:01 1993:01 1994:01 Price of Mexican Equities Mexican Peso-Dollar Exchange Rate IPC local currency index Pesos per U.S. dollar 1995:01 1996:01 8.--------------------~ 3,500 3,000 2,500 2,000 1,500 1,000 3 500 P989:01 1000:01 1991:01 1002:01 1993:01 1994:01 1995:01 L~-~-✓-- ------__.......,...,,...,..,~_ - - - - - 1991:01 1996:01 Stripped Spreads on Mexican Brady Bonds 1992:01 1993:01 1994:01 1996:01 1996:01 Monthly Price Volatility of Mexican Stocks Percent per annum premium over comparable Treasuries Annualized, based on dally price changes within the month 10.----------------------, 30 P989:01 1990:01 1991:01 1992:01 1993:01 1994:01 1995:01 40 -------------------- 1990:01 1991:01 1982:01 199:101 1994:01 1995:01 1996:01 1996:01 1996:01 Mexican Price-to-Earnings Ratio Projections based on 1/B/E/S survey nollar Indices 70030 1,s!lb MSCI world index ----------------------- 1996:01 Prices of Foreign Stocks 1,400 60 - - - - - - - - - - - _,.. 1,300 1,200 1,000 000 Right scale "°\'-. ......0_1_11_...,..,_o"',-,.. ..,..,:o"',....,,"'_':-;-o,~,;;-~_"".'o,~,..,.... ~,o"',-,,...~•"'o"',-,,._~o, 200 feae:01 1990:01 1991:01 1992:01 1993:01 1994;01 Chart 2. Monthly Net Foreign Purchases of Mexican Equities January 1989-March 1996 Millions of dollars 4,000 , - - - - - - - - - - - - - - - - - - - - - - - - - - , Total 3,000 - --- - - - - ----- - ------- - - - - ----- - - --- - - - - - - -· - - - -- - ---- 2,000 Publicized issues 1,000 -1,000 Te mex I and II L..--1-.U..U..U..U..U..U..U..U..U..U..U..U..U..U..U..U...........................................................u.J.J..u.J.JU...U..U..U..U...U..U...U..U......., 1989:01 1989:07 1990:01 1990:07 1991:01 1991:07 1992:01 1992:07 1993:01 1993:07 1994:01 1994:07 1995:01 1995:07 1996:01 Source: Banco de Mexico, International Financing Review, Bank of New York, Bolsa Mexicana de Valores and authors' estimates. Table I. Testable lmpllcatione of Selected Alternative Hypothaala to Account for Corralation between Equity lnflo- and Returns Hypothesis Price changes correlated Price changes correlated with contemporaneous surprise inflows with lagged surprise inflows after controlling for other factors Correlation between flows and returns higher for weekly retums at causing returns the beginniilg of current month and end of previous month No Yes No Yes, negative correlation Yes No Omitted variables • No No No Positive feedback No Yes Yes Base-broadening Price pressure Table IV. Foreign Net Purchases of Mexican Equities billions of U.S. dollars i~ijt(iM Total inflow of which: from the U.S. Publicized placements 1/ Other inflows MemQra □dum 0.5 2.0 6.3 4.8 10.7 4.1 0.0 0.5 1.1 0.9 2.1 2.8 5. 1 1.4 .0.2 0.3 0.5 2.0 3.5 2.8 3.0 1.7 2.8 7.9 1.7 2.3 0.0 0.5 0.1 0.8 0.8 4.1 18.5 28.7 54.6 34.4 24.5 28.3 (3.6) (12.5) (18.3) (20.8) (27.2) (26.5) (27.0) (27.7) 3.4 3.7 7.6 14.1 45.6 34.9 22.0 item; Foreign holdings of Mexican stocks at current market value (in percent of market capitalization) Total portfolio equity flows to developing countries 2/ Sources: Banco de Mexico, U.S. Treasury Bulletin, Bolsa Mexicana de Valores, International Financing Review, World Bank. 1/ Comprises international equity placements identified in Table Ill. 2/ World Bank estimates. Table V. Regressions of Mexican Stock Returns on Net Foreign Purchases of Mexican Equities · Dependent variable Sample period C(O) C(1) Constant Total net foreign purchases 4/ Excess return on Mexican stocks 2/ 2 Peso return on Me.xican stocks 1/ 1 89:02-96.03 Dollar return on Mexican stocks 3/ 3 89:02-96.03 · 89:02-96.03 0.0109 -0.0132 -0.0037 (0.9248) (-1.1314) (-0.2585) 5.8657 ••• 6.1135 ••• 6.4767 (4.6578) (4.6674) (4.3511) Adjusted R2 0.1421 0.1616 0.1282 Durbin-Watson 1.8499 1.9446 1.6305 ••• The regressions are estimated by ordinary least squares with heteroskedasticity consistent covariance. The figures.in parentheses are I-statistics. Significance at the 15, 10, 5 and 1 percent levels are indicated by the symbols#,•,••,•••, respectively. i 1/ Percent change in the IPC index in pesos from the end of the previous month. 2/ Percent change in the IPC index in pesos from the end of the previous month minus Cetes interest rate at the end of the month t-1. 3/ Percent change in the IPC index converted to dollars at the official exchange rate, from the end of the previous month. 4/ Total foreign net purchases in month t divided by market capitalization at the end of month t-1. Table VI. Forecasting Equations for Net Foreign Purchases of Mexican Equities Dependent variable Sample period C(0) C(1) C(2) C(3) C(4) C(5) Constant Unpubliciz.ed net foreign purchases(- 1) Unpublicized net foreign purchases(-2) Publicized net foreign purchases Publicized net foreign purchases (• 1) Publicized net foreign purchases (·2) Unpublicized net foreign Unpublicized Unpublicized eurchases net foreign eurchases 1 net foreign eurchases 2 3 Unpublicized net foreign eurchases 4 89·03-96 03 89:03-96 03 89·03-96 03 ... 89·03·96 03 ... (4.2835) 0.2637 •• (J.3701) 0.4476 (3.4280) ••• (3.9902) 0.1981 0.4156 ••• ... (J.6657) # (1.6574) ·0.0635 0.0013 0.2072 0.0012 (3.3196) 0.4309 (3.8085) • (1.7613) 0.1624 0.0015 0.5162 ... ••• (5.2342) # (I.5906) # (-1.5231) ·0.0786 (-1.6815) • -0.0836 • (-1.7162) ·0.0008 (-0.0207) Adjusted R2 0.2637 0.2745 0.2708 0.2602 Durbin-Wats on 1.9634 1.9679 1.9718 2.1461 Akalke Information criterion ·11.5155 -11.5524 ·]],5586 ·11.5555 Schwartz criterion -11.3431 -11.4375 ·11.4724 ·]],498Q Note: The regressions are estimated by OLS with heteroskedasticity consistent covariance. The figures in parentheses are t-statistics. Lagged values of variables are indicated (-n), where n is the number of months. Significance at the 15, 10 ,5, and 1 percent level are indicated by the symbols#,* ,**, and***, respectively. The variables in these regressions are all scaled by the capitalization of the Mexican stock market at the end of the previous period. Table VII. Regressions of Mexican Stock Returns on Expected and Surprise Net Foreign Purchases of Mexican Equities Dependent variable Sample period C(0) C(1) C(2) C(3) C(4) Peso return on Mexican Peso return on Mexican Peso return on Mexican stocks 1/ 1 Peso return on Mexican stocks 1/ 2 stocks 1/ 3 stocks 1/ 4 89:03-96 03 89'03-96 03 Constant Total net foreign purchases 0.0115 0.0366 (0.9489) (3.8639) 5.8180 12.9081 ... (4.3589) Expected unpublicized net foreign purchases 3/ ...,;,,. Publicized net foreign purchases 4/ 89;03-96,03 0.0227 0.0280 (1.2739) (1,5054) ••• (4.5983) Surprise unpublicized net foreign purchases 2/ ... 89·03-96,03 13.7001 ••• (5.0303) 2.7259 2.2984 (0.5008) (0.4200) 4.2858 ••• (3.4392) C(5) ••• 14.1412 (5.0164) -1.0257 (-0.4164) Telmex Dummy 5/ ••• 0.2713 (2,6974) HllPOtheses tests Probability Values C(3),C(4)=0 4.13% •• 88.79% C(2)=C(3)=C(4) 0.17% ••• 0.03% Adjusted R2 0.1386 0.1597 0.2012 0.2074 Durbin-Watson 1.8364 1.9553 1.9093 1.8833 ••• The regressions are estimated by ordinary least squares with heteroskedasticity consistent covariance. The frgures In parentheses are I-statistics. Significance at the 15. 10. 5 and 1 percent levels are indicated by the symbols#, •, ••, •••, respectively. . · , The indicated probability values for hypothesis c(3)=c(4)=0 are for likelihood ratio tests of the indicated null hypothesis; the test statistics have a chi-squared distribution. The probability values for the hypothesis c(2)=c(3)=c(4) is from Wald test, which also has a chi~squared distribution. 1/ Percent change in the Mexican IPC index in pesos from the end of the previous month. 2/ The residual values of the estimating equation for unpublicized flows •• see column 4 of Table VI. 3/ The fitted values of the estimating equation for unpublicized flows·· see column 4 of Table VI. 4/ Scaled by market capitalization at the end of month 1-1. 5/,This variable equals 1 for the May 1991 Telmex offering. and zero otherwise. Table VIII. Test of the Price Pressure Hypothesis: Regression Tests for Price Reversals Dependent variable Sample period C(O) C(1) C(2) Constant Surprise unpublicized net foreign purchases 2/ Surprise unpublicized net foreign purchases (-1) 2/ C(3) Surprise unpublicized net foreign purchases (-2) 2/ C(4) Surprise unpublicized net foreign purchases (-3) 2/ Peso return on Mexican stocks 1/ 1 Peso return on Mexican stocks 1/ 2 Peso return on Mexican stocks 1/ 3 ~03-96,03 89:03-96.03 /3Mg:.96,_03 0.0366 ••• (3.8639) 12.9081 (4.3589) 0.0335 ••• (3.4958) ••• 13.5652 0.0362 ••• (3.8264) ••• 13.2151 (4.6072) (4.6373) 5.1161 5.6317 (1.4334) (1.6416) ••• # -1.4004 (-0.4627) 1.8921 (0.6352) Hypotheses tests Probability Values C(2),C(3),C(4)=0 29.34% Adjusted R2 0.1597 0.1716 0.1789 Durbin-Watson 1.9553 2.0061 1.9436 The regressions are estimated by ordinary least squares with heteroskedasticity consistent covariance. The figures in parentheses are t-statistics. Significance at the 15, 10, 5 and 1 percent levels are indicated by the symbols# ,',••,'", respectively. The indicated probability value is for a likelihood ratio test of the indicated null hypothesis; the test statistic has a chi-squared distribution. 1/ Percent change in the IPC index in pesos from the end of the previous month. 2/ The residual values of the estimating equation for unpublicized flows -see column 4 of Table VI. Table IX. Regressions of Mexican Stock Returns on Additional Explanatory Variables Dependent variable Peso return on Mexican Dollar return stocks 1/ stocks 1/ 2 on Mexican 1 Semple period C(O) Constant aa·o1-a6 aa 0.0457 (2.3671) C( ) P~rcent change in world stock 1 pnces 2/ C( ) P~rcent chang~ in U.S. stock 2 prices 3/ C( ) Mexican bill rate at the end of 3 the previous month 4/ C( ) Ch~nge in Mexican bill rate 4 dunng month C(S) Adjusted percent decrease in Brady bond stripped yields 5/ C(G) Percent growth in aggregate earnings forecasts 6/ 0.6893 .. ... aa·tu-aG oa 0.0417 (2.1302) 0.7025 aa·o:1-aa oa aa·cu-ae OJ EUHU-96 03 0.0298 0.0091 0.0295 (3.2600) (0.4318) (3.2847) •• ... (3.6657) (3.0993) 0.2868 0.2689 0.3721 (1.4007) {1.3248) (1.1337) 0.0503 0.0443 0.0948 (1.0058) (0.8905) (1.2568) ·0.1184 -0.1192 ·0.4599 (-0.6763) (•0.6686) 0.3603 0.4071 (2.8364) (3.1378) ... 0.2971 0.0950 .. (-1.52S0) • 0.6507 (S.8799) ... 0.3547 (4.0423) (3.6464) ·0.1888 -0.1690 # (•2.8989) {•2.4987) -0.3035 -0.5873 ... (-3.5454) 0.1026 (1.0681) (-4.4952) Adjusted R2 0.5542 0.6846 0.1742 0.0881 0.267367 0.2863 Durbin-Watson 1.8728 1.8752 1.8114 1.9550 2.0593 1.8442 C(7) Stock price volatility 7/ Percent change in the C(B) dollar/peso exchange rate ••• (4.6811) 0.7596 (3.5791) 0.3322 ... 89·01·96 OJ {4.2231) The regressions are estimated by ordinary least squares with heteroskedasticlty consistent covariance. The figures in parentheses are t-statistics. Significance at the 15, 10, 5 and 1 percent levels are indicated by the symbols#,•. 0 , . . . , respectively. 1/ Percent change In the IPC index in pesos trom the end of the previous month. 2/ Percent change in the Morgan Stanley world dollar price index from the end of the previous month. 3/ Percent change in the S&P500 price Index from the end of the previous month. 4/ The last weekly auction rate on 28-day Cates during the" previous month; expressed as an annual rate. 5/ Stripped yields adjusted to remove the impact of change in U.S. treasury bond yields. For further details please see Section II of the text. 6/ See Section II of the text for an explanation of sources arid computation. 7/ Annualized stock price volatility calculated from daily returns during month t. ... Table X. Test of the Omitted Variable Hypothesis: Regress ions of Mexican Stock Returns on Surprise Net Foreign Purchases and Additional Variables Dependent variable Sample. period C(0) Consta nt C(l) Surprise unpublicized net foreign purchases 2/ C( ) Percent change in world stock 2 pnces 3/ Peso return Dollar return stocks 1/ (1 Peso return on Mexican stocks 1/ 2 89·03-96.03 89"03-96.03 89"03-96 03 on Mexican 0.0366 (3.8639) 12.9081 (4.3589) ... ... 0.0454 (2.5720) 7.9793 on Mexican stocks 1/ 3 •• ... (4,0868) 0.5619 Dollar return on Mexican 0.0251 stocks 1/ 4 .. (2.2328) 12.6260 ••• (4.0481) ••• C( ) Mexican bill rate at the end of the. 4 previou s month 5/ C( ) Change in Mexican bill rate 5 during month C( ) Adjusted percent decrease in 6 Brady bond stripped yiel_ds 6/ C( ) Percent growth in aggregate 7 earning s forecasts 7/ 0.1597 Durbin-Watson 1.9553 (3.6989) (3.1940) 0.3283 (1.5619) 0.0507 0.0462 (1.0188) (0.8782) ·0.156 7 -0.1562 (-0.9416) (·0.8989) 0.2319 • (1.7061) ... ... -0.1780 Adjusted R2 ... ... 7.4312 (1.6735) (4.4580) Percent change in the C(9) dollar/peso exchange rate •• 0.3547 0.3500 C(B) Stock price volatility 8/ 0.0417 . (2.1827) 0.5851 (3.0622) C( ) Percent change in U.S. stock 3 pnces 4/ 89:03-96.03 0.2834 (2.0202) 0.3138 (3.9754) -0.1611 (-2.7694) (-2.3769) 0.0224 ·0.6621 (0.2306) (-4.7369) 0.6093 0.1117 0.6977 1.8891 1.6859 1.9229 The regressions are estimated by ordinary least square s with heteroskedasticity consistent covariance. The figures in parentheses are t·statistics. Significance at the 15. 10. 5 and 1 percent levels are Indicated by the symbo ls#,•,· ·, ***, respectively. 1/ Percent change in the IPC index in pesos from the end of the previous month. 2/ The residual values of the estimating equation for unpublicized flows• · see column 4 of Table VI. 3/ Percent change in the Morgan Stanley world dollar price index from the end of the previous month. 4/ Percent change In the S&P500 price index from the end of the previous month. 5/ The last weekly auction rate on 28·day Cetes during the previous month; expressed as an annual rate. 6/ Stripped yields adjusted to remove the impact of change in U.S. treasury bond yields. For further details please see Section II of the text. 7/ See Section II of the text for explanation of source s and computation. 8/ Annualized stock price volatility calculated from daily returns during month t. # •• ... .. ••• Table XI. Test of the Positive Feedback Hypothesis: Regression of Surprise Net Purchases on Current and Lagged Weekly Stock Returns Surprise Dependent variable Sample period C(0) Constant unpublicized Surprise unpublicized net foreign purchases 1/ 1 0.0132 2 3 4 89:03-96,03 # (-1.5731) C( ) Monthly peso return on 1 Mexican stocks 2/ unpublicized net foreign purchases 1/ net foreign 89·03-96,03 -0.0005 purchases 1/ Surprise unpublicized net foreign purchases 1/ Surprise 89:03-96 03 89:03-96 03 -0.0004 .-0.0004 -0.0005 (-1.3317) (-1.3317) (-1.5905) # ••• (4.2689) C( ) Monthly peso return on 2 Mexican stocks (• 1) 2/ -0.0001 (-0.0405) C( ) Sum of weekly returns on 3 Mexican stocks 3/ 0,0218 ••• (2.6983) C( ) Week 4 return on Mexican 4 stocks 3/ 0.0218 0.0156 (4.4988) ••• (2.6983) C( ) Week 3 return on Mexican 5 stocks 3/ 0.0128 •• -0.0090 (2.0712) C( ) Week 2 return on Mexican 6 stocks 3/ 0.0203 C(?) Week 1 return on Mexican stocks 3/ C(B) Week 4 return on Mexican stocks (-1) 3/ C( ) Week 3 return on Mexican 9 stocks (-1) 3/ (-0.8882) •• -0.0015 (2.0673) (-0.1131) 0.0103 -0.0115 (0.8905) (-0.9211) 0.0002 0.0002 (0.0169) (0.0169) -0.0015 -0.0015 (-0.2140) (-0.2140) Hypotheses tam ProbabHity 'lllll.lH C(5),C(6),C(7),C(B),C(9)=0 90.41% Adjusted R2 0.1698 0.1289 0.1289 0.1660 Durbin-Watson 2.3273 2.3820 2.3820 2.3221 The regressions are estimated by ordinary least squares with heteroskedasticity consistent covariance. The figures in parentheses are t-statistics. Significance at the 15, 10, 5 and 1 percent levels are indicated by the symbols #, •, ••, •••, respectively. The indicated probability value is for a likelihood ratio test of the indicated null hypothesis: the test statistic has a chi-squared distribution. 1/ The residual values of the estimating equation for unpublicized flows •· see column 4 of Table VI. 2/ Percent change in the IPC index in pesos from the end of the previous month. 3/ Percent change in the Morgan Stanley local currency price index for Mexican stocks. Week 4 refers to return over the last 7 calendar days of the month, Week 3 to the 7 days before Week 4, Week 1 to the first seven days of the month, and Week 2 to the second 7 days. ••• FEDERAL RESERVE BANK OF NEW YORK RESEARCH PAPERS 1996 The following papers were written by economists at the Federal Reserve Bank of New York either alone or in collaboration with outside economists. Single copies of up to six papers are available upon request from the Public Information Department, Federal Reserve Bank of New York, 33 Liberty Street, New York, NY 10045-0001 (212) 720-6134. 9601. Bartolini, Leonardo, and Gordon M. Bodnar. "Are Exchange Rates Excessively Volatile? And What Does 'Excessively Volatile' Mean, Anyway?" January 1996. 9602. Lopez, Jose A. "Exchange Rate Cointegration across Central Bank Regime Shifts." January 1996. 9603. Wenninger, John, and Daniel Orlow. "Consumer Payments over Open Computer Networks." March 1996. 9604. Groshen, Erica L. "American Employer Salary Surveys and Labor Economics Research: Issues and Contributions." March 1996. 9605. Uctum, Merih. "European Integration and Asymmetry in the EMS." April 1996. 9606. de Kock, Gabriel S. P., and Tanya E. Ghaleb. "Has the Cost of Fighting Inflation Fallen?" April 1996. 9607. de Kock, Gabriel S. P., and Tania Nadal-Vicens. "Capacity Utilization-Inflation Linkages: A Cross-Country Analysis." April 1996. 9608. Cantor, Richard, and Frank Packer. "Determinants and Impacts of Sovereign Credit Ratings." April 1996. 9609. Estrella, Arturo, and Frederic S. Mishkin. "Predicting U.S. Recessions: Financial Variables as Leading Indicators." May 1996. 9610. Antzoulatos, Angelos A. "Capital Flows & Current Account Deficits in the 1990s: Why Did Latin American & East Asian Countries Respond Differently?" May 1996. • 9611. Locke, Peter R., Asani Sarkar and Lifan Wu. "Did the Good Guys Lose? Heterogeneous Traders and Regulatory Restrictions on Dual Trading." May 1996. 9612. Locke, Peter R. and Asani Sarkar. "Volatility and Liquidity in Futures Markets." May 1996. 9613. Gong, Frank F., and Eli M. Remolona. "Two Factors along the Yield Curve." June 1996. 9614. Harris, EthanS. and.Clara Vega. "What Do Chain Store Sales Tell Us About Consumer Spending?" June 1996. 9615. Uctum, Merih, and Michael Wickens. "Debt and Deficit Ceilings, and Sustainability of Fiscal Policies: An Intertemporal Analysis." June 1996. 9616. Uctum, Merih, and Michael Aglietta. "Europe and the Maastricht Challenge." June 1996. 9617. Laster, David, Paul Bennett, and In Sun Geoum. "Rational Bias in Macroeconomic Forecasts." July 1996. 9618. Mahoney James M., Chamu Sundaramurthy, and Joseph T. Mahoney. "The Effects of Corporate Antitakeover Provisions on Long-Term Investment: Empirical Evidence. July 1996. 9619. Gong, Frank F., and Eli M. Remolona. "A Three-Factor Econometric Model of the U.S. Term Structure." July 1996. 9620. Nolle, Daniel E., and Rama Seth. "Do Banks Follow Their Customers Abroad? 9621. McCarthy, Jonathan, and Charles Steindel. "The Relative Importance of National and Regional Factors in the New York Metropolitan Economy." July 1996. 9622. Peristiani, Stavros, Paul Bennett, Gordon Monsen, Richard Peach and Jonathan Raiff. "Effects of Household Creditworthiness on Mortgage Refinancings." August 1996. 9623. Peristiani, Stavros. "Do Mergers Improve the X-Efficiency and Scale Efficiency of U.S. Banks? Evidence from the 1980s." August 1996. 9624. Ludvigson, Sydney. "Consumption and Credit: A Model of Time-Varying Liquidity Constraints." August 1996. 9625. Ludvigson, Sydney. "The Channel of Monetary Transmission to Demand: Evidence from the Market for Automobile Credit." August 1996. 9626. Sobol, Dorothy M. "Central and Eastern Europe: Financial Markets and Private Capital Flows." August 1996. 9627. Evans, Joan, and James M. Mahoney. "The Effects of Daily Price Limits on Cotton Futures and Options Trading." August 1996. 9628. Molyneux, Philip, and Rama Seth. "Foreign Banks, Profits, and Commercial Credit Extension in the United States." August 1996. 9629. Cantor, Richard, and Robert Driskill. "Can a Fiscal Contraction Strengthen a Currency? Some Doubts about Conventional Mundell-Fleming Results." August 1996. 9630. Jayaratne, Jith, and Philip E. Strahan. "Entry Restrictions, Industry Evolution, and Dynamic Efficiency: Evidence from Commercial Banking." August 1996. 9631. Dziwura, Joseph R., and Eric M. Green. "Interest Rate Expectations and the Shape of the Yield Curve." September 1996. 9632. Brewer, Elijah III, and Marc R. Saidenberg. "Franchise Value, Ownership Structure, and Risk at Savings Institutions." September 1996. 9633. Fleming, Michael J., and Eli M. Remolona. "Price Formation and Liquidity in the U.S. Treasuries Market: Evidence from Intraday Patterns around Announcements." October 1996. 9634. Orlow, Daniel K., Lawrence J. Radecki, and John Wenninger. "Ongoing Restructuring of Retail Banking." November 1996. To obtain more information about the Bank's Research Papers series and other publications and papers, visit our site on the World Wide Web (http://www.ny.frb.org). From the research publications page, you can view abstracts for Research Papers and Staff Reports and order the full-length, hard copy versions of them electronically. Interested readers can also view, download, and print any edition in the Current Issues in Economics and Finance series, as well as articles from the Economic Policy Review.