Monetary Policy and Reaching for Income by Daniel, Garlappi and Xiao. Discussant: Annette Vissing-Jorgensen, UC Berkeley.

Transcription

1 Monetary Policy and Reaching for Income by Daniel, Garlappi and Xiao Discussant: Annette Vissing-Jorgensen, UC Berkeley April 28, 2018

2 Findings: Following lower Fed funds rate (over 3 years). 1) Mutual funds, : retail investors move money into equity mutual funds, especially high DY funds. Related result in Hau and Lai (JFE 2016) on European data should be cited. Impact of monetary policy on fund flows to equity funds and money market funds using real rate variation from cross-eu differences in inflation rates , 8 EU countries.

3 2) Individual stocks, : retail investors buy high dividend-yield stocks. No substantial effect for stocks in general. Stronger effect for retirees and withdrawers. Robust to using deposit rates instead of Fed funds rate, which allows inclusion of time fixed effects. 3) Stock returns, : high dividend-yield stocks have avg. alphas that are 39 bps/month higher than the avg. alphas of low-dividend yield stocks. 56 bps/month higher in dif-in-dif.

4 Argument: Time-inconsistent retail investors wanting to avoid overconsuming rely on consumption from dividends and interest. This leads them to buy higher dividendyield stocks following declines in interest rates. The market doesn t expect that retail investors will behave like this. As they gradually reallocate to high-dividend yield stocks, prices of these stocks go up (relative to low-dividend stocks). Observation: This is quite different from reaching for yield Reaching for yield: Taking more risk to gain extra expected return. Reaching for income: Taking less risk to get more of your return from dividends (high dividend-yield stocks are low beta, high value and low volatility.)

5 General impression: The portfolio choice analysis is quite convincing The asset pricing analysis is not (yet?)

6 Comment 1 (portfolio choice analysis): Control for beta, value. Add bond analysis. Check: In the portfolio choice analysis Are people reaching for income or something else? High dividend yield stocks are low beta and value (and low volatility). Include interaction terms ΔFFR*(beta) and ΔFFR*(B/M) in portfolio choice analysis

7

8 And study bonds: The theory predicts that a lower riskless rate should lead to 1) Reallocation within stocks to those with high Dividend/Price: TESTED 2) Reallocation within bonds to those with high Coupons/Price: NOT TESTED Test in both mutual fund data and Odean data. Test asset pricing implications: Do bonds with high Coupon/Price ( high current yield ) outperform (on a risk-adjusted basis) following lower Fed funds rates? One nice aspect: You get somewhat exogenous variation in coupon rates Coupons tend to be based on what interest rates (and credit risk) were at the time of issue. You can distinguish between current yield and yield.

9 Total assets of High Current Yield funds in the CRSP mutual fund database: jul jan jul jan jul jan2012 Total Institutional Retail

10 Increases line up quite well with declines in Fed funds rate: Effective Fed funds rate, monthly avg. 36-month change

11 Comment 2 (asset pricing): If you control for realized inflation news, the asset pricing result is a lot weaker The story is about slow retail investor reallocation that the market didn t expect. It shouldn t matter whether one controls for realized inflation news, but it does. Why? High dividend-yield stocks benefit from low inflation (news). In the data, after Fed funds rates fell, inflation was lower than expected. Controlling for realized inflation news, changes in Fed funds rate have little predictive power for 10-1 return. Economics? Perhaps related to differential stickiness of costs or prices (matters for returns, see various Weber papers).

12

13

14

15 You can see the sensitivity to inflation news in the cumulative return graph around 1980:

16 Year 10 1, annual return Strategy, annual return Annual inflation Fraction of months strategy is long does poorly in as inflation is surprisingly high 10-1 does well in as inflation is surprisingly low (Volcker) Strategy is mostly short 10-1 so does the opposite.

17 Sensitivity of high dividend-yield stock returns to inflation may also matter for portfolio choice analysis: - After lower short rates, investors buy high DY stocks - Are they reaching for income, or do they see low inflation and think they can get into high DY stocks before their prices fully reflect this? - Retail investors may not understand that by the time they buy the inflation news is probably already in the price.

18 Comment 3 (asset pricing): Why no analysis of days with FOMC news? Normally in a paper arguing that monetary policy drives certain asset returns you focus on days with monetary policy news: FOMC announcement days That approach is not used here, because it takes years for retail investors to react to the monetary policy news. So everything is done with monthly data. But do markets really only react with a long lag? As long as markets have some idea that lower rates will lead investors to reach for income, then you should see high returns on the high DY portfolios on days with accommodating monetary policy news. It doesn t matter if investors later reach more than we thought, we just need some understanding that they reach to get better identification.

19 , FOMC announcement days, using daily dividend-yield deciles returns I constructed from CRSP: Doesn t work: Negative monetary policy rate surprises do not benefit Col 1, 2: Usual regression of market excess return on monetary policy surprise Col 3, 4: Regression of 10-1 return on monetary policy surprise Col 2, 4 drops 2008 which has an outlier.

20 Comment 4: How does monetary policy move risk premia? Current paper: Low rates Lower risk premium on high DY stocks. Reaching for yield: Low rates Lower risk premia in general. But: 1) Explanatory power of policy surprises for stocks on FOMC announcement days is small. We saw that for both 10-1 and market. 2) Lucca and Moench: Stocks on avg. do well on FOMC announcement days despite no avg. reduction in Fed funds rates/futures (Lucca and Moench). 3) Cieslak, Morse and Vissing-Jorgensen: Stocks on avg. do well in even weeks in FOMC cycle time With only small reductions in Fed funds futures. Following market declines: Stocks do well in even weeks despite no reduction in Fed funds futures.

21 How is this possible? Can the Fed directly affect risk premia? Or are Fed funds futures (and other short bond yields) not good measures of policy surprises? A possible resolution: 2) is correct whenever policy is effective. A promise of future accommodation should lower futures Unless the promise lowers the probability of bad future economic state.

22 Example: Suppose the Taylor rule was like this: State of economy in 1 year Bad Good Probability Unemployment 8% 4% Fed funds target 2% 6% Expected Fed funds target: 4%

23 Then suppose the Fed says: If the economy gets bad, we will lower to 0%. Without any effect on probability of bad and good state: State of economy in 1 year Bad Good Probability Unemployment 8% 4% Fed funds target 0% 6% Expected Fed funds target: 3% But if policy shock lowers probability of bad state: State of economy in 1 year Bad Good Probability Unemployment 8% 4% Fed funds target 0% 6% Expected Fed funds target: 4% An accommodating policy shock may not move the expected Fed funds target down!

24 So how do we measure monetary policy shocks? Ideally you d ask policy makers to state what the appropriate policy rate is for various values of unemployment and inflation: Conditional dot plots Or you need a derivative that pays off as a function of both unemployment and the Fed funds rate (or both inflation and the Fed funds rate).