安蒂·伊尔曼宁 - 理解收益预期（第七季第21集）

如果我们采用基于估值的资本市场假设，并假设它们是以客观可行且理性的方式构建的，然后将它们输入到一个受约束的均值-方差优化模型中，我们通常会得到一个与任何你见过的机构政策组合都截然不同的投资组合。

If we take valuation based capital market assumptions and we give them the benefit of the doubt that they are developed in an objectively feasible and rational manner and we plug them into a constrained mean variance optimization, we usually end up with a portfolio that looks absolutely nothing like any institutional policy portfolio that you've ever looked at.

我曾经做过一个名为theweirdportfolio.com的副项目，我取了——我不会说出名字——但那是一家非常庞大且复杂的机构，我会使用他们每年发布的资本市场假设。

I actually used to run as a side project, a website called theweirdportfolio.com where I took, I won't mention the name, but it was a very large sophisticated institution and I would take their capital market assumptions that they published every year.

我甚至不会把那些奇怪的资产类别放进去。

And I would say, I'm just not even gonna put the weird asset classes in here.

我只使用股票和债券，并以60/40组合为目标，利用他们的相关系数矩阵和协方差矩阵来匹配60/40组合的波动率特征。

I'm just gonna use the stocks and bonds and I'll target the same variance as a sixty forty using their correlation matrix and covariance matrix to come up with the sixty forty volatility profile.

结果却与任何机构实际可能持有的组合相差甚远。

And the result was just so wildly different than something that any institution would feasibly hold.

这正是你之前在对话中提到过的问题。

And so this is something you've mentioned earlier in the conversation.

但你如何调和这种矛盾：这些假设在理论上是客观可行的市场预期回报，但由此推导出的投资组合却无人实际持有？

But how do you reconcile this disconnect that these are in theory, the objectively feasible market expected returns, and yet the implied portfolio is something that nobody actually holds.

是的，没错。

Yeah, yeah.

你提到我们可能遗漏了一些有趣的另类资产，这些似乎更多与投资者的约束条件等相关。

You mentioned that there could be some interesting alternatives that we leave out, which are, I think, more related to investor constraints and so on.

但总体而言，我认为机构——根据数据，我知道机构在纸面上，基于他们的资本市场假设等，看起来比散户投资者更具反向思维，而散户则更倾向于外推。

But overall, I do think that institute well, I know from data that institutions on paper with their CMAs and so on, they look more the contrarian objective than retail investors do, which are much more extrapolatable.

但正如你所说，他们的实际投资组合却并未完全遵循这种逻辑。

But then, like you say, their actual portfolios don't follow quite this.

一个很好的例子就是美国市场与非美国市场的对比。

Good case in point is The US versus non US.

再比如，任何基于收益率或估值的估算都表明，在过去十五年里，你应该对美国市场保持低配，但许多机构并没有这样做，而他们对此感觉非常好。

Like, again, any yield based or valuation based estimate would tell for the last fifteen years that you should be underweighting US, and many institutions didn't, and they feel very, very good about it.

所以我认为，机构在自我审视时，普遍倾向于认为自己是耐心的反向投资者。

So I think there is this broader thing that institutions do like to think of themselves as when they look in the mirror, they see a patient contrarian investor.

但当他们真正操作投资组合时，却并不总是具备这种耐心，也不总是表现出反向思维。

And then when they actually act with their portfolios, there's not always the patience and there isn't the contrarian.

而有时，这种做法反而对他们有利。

And sometimes it works in their benefit.

在过去的十年里，尤其是在美国股市中，不采取逆向策略是有利的。

And in the last decade, certainly with US equities, it's been good not to be contrarian.

但我认为这是一个更广泛的现象，即 extrapolating 的顺周期倾向或诱惑，也存在于机构层面，我认为这可能是最重要的解释。

But I do think that that's a broader point that, I don't know, the procyclic tendencies of extrapolating they or their temptations, they also happen at the institutional levels and I think that's probably the most important explanation.

基于估值的资本市场假设是否与有效市场假说相矛盾？

Do valuation based capital market assumptions contradict the efficient market hypothesis?

换句话说，将估值纳入资本市场假设是否只是一种伪复杂的市场择时形式？

In other words, is including valuations in a capital market assumption just a pseudo sophisticated form of market timing?

是的，我认为这样说很公平。

Yeah, I think it's fair.

我不认为这太复杂，因为你只关注大约十年的期限，这意味着你只关心收益率和估值。

I wouldn't call it too sophisticated because you are only looking at sort of ten year horizons, which means that you only care about yields and valuations.

你忽略了短期市场择时信号、动量、经济趋势、情绪流动指标等。

You ignore short term market timing signals, momentum, economic trends, sentiment flow indicators, etcetera.

因此，这是用于市场择时的最佳工具。

And so it's sort of the best possible thing to use for market timing.

而且我们确实知道，过去十年美国股市的表现非常糟糕，这一点我们经常提到。

And then we do know that the record is what we talk often now about this fact that it really did badly with US equities in the last ten years.

但同样地，从更长的历史来看，它的表现优于抛硬币，也优于后视镜式的判断。

But again, taking longer history, it is better than coin flip, it's better than rearview mirror.

是否将这些整体证据视为积极的，这取决于我们的选择。

It's a choice whether we take that overall overall evidence as positive.

但我认为，将其视为某种形式的市场择时是公平的。

But I think it is fair to think of it as some kind of market timing.

如果我们对基于估值的资本市场假设给予善意，认为它们确实是理性客观的，那么如果所有投资者都理性地使用这些假设来构建投资组合，市场会是什么样子？

If we give the benefit of the doubt to valuation based capital market assumptions that they are truly rationally objective, what do markets look like if we take it to the extreme and say that every investor behaved rationally and use them to form their portfolio?

在这种情况下，价格错配会消失，还是收益的分散性最终仍不可避免？

Is this a case where mispricings vanish, or is dispersion among returns ultimately still inevitable?

是的。

Yeah.

这有点像行内话，但我提醒一下，这里总有一个风险故事，这个风险故事可能是我们正处于2009年全球金融危机时期，所有人都在这个环境中变得更加风险厌恶。

This is a bit inside baseball here, but but I remind that there's always a risk story here, and the risk story could be that we are in GFC 2009, everybody's getting in this environment more risk averse.

我们都在要求为风险资产组合获得更高的预期回报，但所有人不可能同时涌向退出。

We are all demanding higher required returns for our risky asset portfolios, and yet we can't all rush to the exit at the same time.

因此，这是一种所有人拥有共同预期，却无法以相同方式行动的情况，因为每个买家都需要有卖家。

So that would be a situation where everybody would have common expectations and yet we couldn't all act in the same way because for every buyer there needs to be a seller.

一个合乎逻辑的想法是，某些投资者，比如通常具有长期视角的机构，可能充当逆向市场择时者，在这种情况下承担更多风险——这正是当那些更具动态风险规避倾向的投资者在那时几乎于底部抛售时的初衷。

Like one logical idea would be maybe that situation that some investors typically you can think of long horizon institutions, they could be sort of the contrarian market timers and take more risk in that situation that was sort of the idea when those more dynamically risk averse investors would be basically selling at the bottom at that time.

但这里更广泛的观点是，我们所有人拥有相同预期是可行的，但我们不可能所有人的主动操作都一致，因为每个买家都对应一个卖家，这就是我们在2010年第二点一会议上讨论过的那则故事。

But the broader point here is that it's feasible to think that we would all have the same expectations, but we can't have all same active deals for every buyer, that's the seller, that's who is on the other side story that we talked about in twenty ten's in two point one conference as you recall.

在我迄今为止阅读的这一系列中，我认为最有趣的一个主张是：股票投资者倾向于外推，而债券投资者则倾向于预期均值回归，你提出‘显著性’是造成这种差异的驱动因素。

One of the claims that I thought was the most interesting of the series as I've read it to date, recognizing there's more papers to come, but was this idea that equity investors extrapolate while bond investors tend to expect meaner version and you suggest this idea of salience as being the driver behind this difference.

我希望你能进一步阐释这一概念，以及它为何在股票和债券投资者的预期之间造成如此显著的差异。

Was hoping you could unpack this concept and why it creates such a stark difference in the expectations of equity and bond investors.

是的。

Yeah.

让我们暂时退一步，澄清一下这一证据。

Then let's step back for a moment to clarify that evidence.

我们已经讨论过这一点：至少散户股票投资者和股票分析师有着更为乐观的预期。

So we've already talked about this point that equity, at least retail equity investors and equity analysts have got these extra positive expectations.

在行情好的时候，他们对回报或增长等的预期会变得更加乐观。

After good times, they get even more bullish on returns or growth or so on.

而当我们观察利率投资者时，却看到几乎完全相反的情况。

Whereas when we look at rates investors, we see pretty much the opposite.

如果我们经历了一个收益率降至历史低位的十年，比如20年，那么利率投资者的正常反应是预期某种回归常态。

That if we get a decade where yields are falling to all time low levels, like think of 20, well, the normal behavior by rates investors is to expect some normalization.

因此，你基本上会看到一条向上倾斜的曲线，这似乎反映了均值回归的利率预期。

So you have a basically upward sloping curve, which seems to be a case of mean reverting rate expectations.

然后当我们回顾2022年发生的情况时，一旦利率上升，收益率曲线就自动出现了倒挂。

And then when we think of what happened in '22, once rates went to higher levels, we got inversion pretty automatically.

我认为，这最好被理解为均值回归的利率预期，而不是某种期限溢价的故事。

And again, I think it's best understood not as some term premium story, but mean reverting rate expectations.

人们大致会参考过去十年的平均收益率，并认为我们会回归到那个水平。

People roughly look at last ten year average yields and think that we are gonna go back towards that.

而且，正如我展示的那样，这一点将在未来的一篇文章中呈现。

And that actually, like I show, that's gonna come in a future piece.

我展示了收益率曲线形态与短期利率水平的去趋势变化之间非常明显的模式或共同变动，这与这种均值回归的利率预期一致。

I show quite nice pattern or co movement between yield curve shape and just the detrend in short rate level, which is consistent with this mean reverting rate expectation.

因此，数据似乎表明，正如我所声称的那样：股票投资者倾向于外推预期，债券投资者则持反向预期，尤其是普通投资者。

So it seems, again, data suggests this is what I claim, extrapolative equity investors, contrarian bond investors, average investors in particular.

我不太确定信贷市场是否处于中间状态。

I don't know about credit somewhere halfway.

顺便说一句，这一点对市场远期利率和经济共识等都成立。

And this is, by the way, true both for market forwards and economic consensus and so on.

无论如何，下一个问题是为什么，我会谈到显著性问题，但首先我们考察的是：这是否仅仅因为历史上我们一直看到这种情况。

Anyway, so the next question is why, and I'll get to that salience, but the first thing we looked at was, okay, could this be just that this makes sense because historically that's what we have seen.

我们通常看到，增长序列倾向于延续，而债券序列则倾向于多年反转。

We tend to see extra growth series tend to have continuation tendencies and bond series tend to have multi year reversal tendencies.

但事实上，历史证据恰恰相反。

But actually historical evidence is pretty much the opposite.

因此，我不能说这是对每个资产类别中发生情况的某种理性反应。

So I can't say that this is somehow a rational reaction to what has happened in each asset class.

于是，我转向了我的直觉想法，即这个显著的故事。

So then I go to my intuitive idea which is this salient story.

这个显著的故事非常简单，就是说，我是个老债券人，我确实更习惯通过收益率的视角来看待这个世界，我们总是引用收益率，这让我更自然地以前瞻性的方式思考股票。

And the salient story is quite simple, just saying that, well, I'm an old bond guy and I certainly think about the world much more through the lens of yields, we quote yields, and that has made it more natural for me to think also about equities in a forward looking sense.

而大多数进入股票领域的人，他们用价格来报价，谈论的是过去一年或过去几年任何股票类投资的表现，这自然赋予了他们一种积极的心态，而不是债券投资者带来的前瞻性反向思维。

Whereas most people who come to equities, they quote them in prices and they talk about last year or last few years performance of any equity type investment, and that naturally gives this type of extra positive mindset to them rather than the forward looking contrarian mindset that bond investors bring to the table.

因此，我认为这确实归结于我们如何报价、如何谈论这些市场这类简单的事情，而这些对市场行为产生了出人意料的深远影响。

So I do think that it really goes back to these kind of simple things of how we are quoting things, how we are talking about those markets, and it has got surprisingly deep impact then on market behavior.

你还指出，在过去长达三四十年的长期债券牛市期间，尤其是在美国，专家和市场表现出的均值回归利率预期事后看来是愚蠢的。

You also note that experts and markets exhibited mean reverting rate expectations that looked foolish ex post during this really long bond bull market we had, particularly in The United States over the last thirty, forty years.

我的问题是，尽管他们在长达三四十年的时间里判断错误，你为何认为这在事前其实是理性的呢？

And my question is, why do you argue that this was actually rational ex ante despite them getting it wrong for three thousand forty years, right?

在某个时刻，你是否会说，好吧，犯错十年可以原谅，但也许你该更新你的模型了。

At a certain point do you say, okay, you're forgiven for getting it wrong for ten years, but maybe you should be updating your models.

也许你的事前模型在当前经济环境下根本就是错误的。

Maybe your ex ante model is just completely incorrect for the regime.

你还能多久一直说某人行为是理性的，却持续在预测上出错？

How long can you continue to say that someone is acting rationally and yet be continuously on the wrong side of the forecast?

是的。

Yeah.

是的。

Yeah.

在我的两本书中，我都有这种直觉，很高兴最近的一些学术研究正式化了我原有的这种直觉。

I had this intuition in both of my books, and it's nice that some recent academic studies have formalized the intuition that I have.

这涉及到对渐进式结构性变化或这些非常持久的走势的理解。

And it is learning about gradual structural changes or these very persistent movements.

这真的很难。

It is really hard.

如果像我们看到的收益率变化那样，比如通胀持续上升三十年然后下降四十年，实际收益率长期下行，或者我们看到的CAPE趋势，这些都很容易让我们出错。

And if something like what we saw with yields, let's say inflation going up thirty years and then falling forty years, and real yields coming down for a long time, or again, CAPE trends that we see, those really trip us easily.

当这些事情发生时，我们会事后诸葛亮地认为，这本应该是可以预见的。

And when those things happen, we get the hindsight that this really should have been foreseeable.

但在任何时间点，我们都触及了新的历史低点，并预测还会进一步跌向更低的水平。

But at any point in time, we were hitting some new all time lows and predicting that we are going to go into even further lows.

这显然与历史证据不一致，因为当时我们已经处于历史低点，而这些序列本身存在一定的均值回归倾向等等。

Certainly wouldn't be consistent with sort of the historical evidence where we are already at all time lows and there has been some mean reversion tendency in those series and so on.

所以我认为我非常理解并同情这种逻辑。

So I think I am very understanding and sympathetic to this type of logic.

一些学术研究显示，当他们建模一个理性投资者，面对近几十年来通胀或实际利率变化的证据时，他们认为，均值回归的预期始终是理性的，但结果却一再、再三、再三地错了。

Then so some academic studies have shown that when they model a rational investor who faces evidence on what's been happening to either inflation or some real rate behavior in recent decades, they said, Well, mean reverting expectations has been the rational thing to do all the time, but it just turned out wrong, wrong, wrong many times.

你知道，债券在2022年表现还不错，但当你放眼更长的历史时，这只是一个微不足道的借口。

You know, with bonds, '22 turned out fine, but it's sort of a small excuse when you take this longer history.

在股票方面，我们很久没有经历过这种估值的均值回归了，而这种回归本应发生。

With equities, we really haven't gotten in a long time, we haven't gotten this, I don't know, due mean reversion in valuations that would happen.

所以我认为，这很好地描述了近几十年来发生的事情。

So I think this is a good descriptive story of what has happened in recent decades.

因此，这让我想到，尽管这些股票分析师的行为确实显得非理性，但我们完全可以讲出一个非常合理的理性故事，这让我们对这些愚蠢的错误少一些嘲笑。

And so it makes me think that while those equity analysts' behaviors, they really are irrational here, we can tell a very plausible rational story, which should make us laugh a little less at the foolish errors.

你知道，我们还展示了这些齿轮持续二十年下跌的图片，而人们却始终预期利率随时会上升，你们什么时候才能吸取教训之类的话。

You know, we also showed these pictures of gears falling for twenty years and people expecting rising rates any point in time, when will you ever learn or something like this.

但同样地，以事后之明来看，这一点我们其实可以很好地解释。

But again, with hindsight, this is something which we can explain pretty well.

我想补充的最后一点是，我们刚才谈到了趋势，但当我们观察趋势内部的情况时，如果只看这些趋势内的货币政策周期，实际上预期表现得相当不错。

And the last point I want to just add here is that now we were talking about the trend, but when we look at within trend, if you look at just monetary policy cycles within those trends, actually expectations were pretty good there.

因此，每当出现政策收紧，比如我们在2022年之前的情况，就会出现收益率曲线倒挂，以及对未来利率下降的预期。

So you had basically whenever you got this policy tightening, such as we had up to '22, then you had basically inverted curve and expectations for falling rates.

当利率处于极低水平、政策极度宽松时，市场则预期会收紧。

When you had very, very low rates, very easy policy, you had tightening expectations.

这些情况通常都得到了验证。

And those things tended to work out fine.

问题在于趋势持续下行、再下行，任何预测共识——也许不是每个预测者，但任何共识序列或市场序列——都未能预测到这一点。

The problem was with the trend keep going down, down, down, which any forecaster consensus, maybe not every forecaster, but any consensus series or market series miss.

在股票市场中，您指出投资者倾向于推断美国的特殊性，尤其是美国科技领域。

In equities, you note that investors tend to extrapolate US exceptionalism, particularly around US tech.

您那篇专门聚焦美国与非美国股票的论文表明，美国与非美国之间的大部分优势实际上是估值驱动的，而非增长驱动的，即使对这些问题进行了调整。

And your paper that focused explicitly on US versus non US equities showed that most of that the edge between US versus non US was actually valuation driven, not growth driven, even adjusting for these issues.

我想知道，您认为投资者为何持续误判这一点？

Curious why you think investors persistently misattribute this.

根据我至少是基于个人观察或实际对话的证据，人们确实认为过去十年美国跑赢国际市场的原因是美国的增长，而不是估值因素。

It's at least my anecdotal evidence or anecdotal realized conversations that people do think that it was US growth that was the reason why US outperformed international and it wasn't a valuation story over the last decade.

我认为这背后存在某种人类的懒惰，因为数据实际上讲述了一个略有不同的故事。

Well, I think there is some human laziness because the numbers again do tell a somewhat different story.

但有一个可以解释的理由是，美国确实一直存在一定的增长优势，即使回溯五十年或一百多年也是如此。

But the one excuse that there is, is that there truly has been some US growth edge, even when we go back fifty or one hundred plus years.

因此，长期以来，美国相对于世界其他地区大约有1%的增长优势，您甚至可以通过非常长期的Dim Sum及其合著者系列数据看到这一点。

So there's something like, let's say, 1% US growth advantage has been there over the rest of the world, and you can see it even with very long run, Dim Sum and his co author series.

我们可以将这种优势归因于美国独特的创业文化、更好的制度等等。

And we can attribute it to exceptionalism from the entrepreneurial culture, better institutions and so on.

或者只是运气好，因为你们没有经历过本土战争、恶性通货膨胀或革命等等。

Or could be just luck that you didn't have war on own soil, hyperinflation revolution and so on.

但无论如何，我们还是应该更多地归功于美国的特殊性，而不是运气。

But anyway, let's give more credit to US exceptionalism than luck.

在金融危机之后，我们获得的远不止那1%，人们迅速将这种优势 extrapolate 到永恒的未来，导致估值不断攀升。

Now post GFC, we just got much more than that 1%, and people have been quickly willing to extrapolate that to the kingdom come or for whatever indefinite future, and that has led to those valuations to grow.

因此，我想给这些过热的想法泼点冷水，提醒大家，比如，金融危机后美国的超额表现确实存在。

And so then I would want to sort of throw some ice to the hot thoughts there by reminding, for example, that US outperformance post GFC has been there.

但人们是否知道，美国在2000年代、1980年代、1970年代实际上表现落后？

But do people know that US actually underperformed in 2000s, in 1980s, 1970s?

历史上有很多十年，美国的表现是落后的。

There have been many decades where US has underperformed.

即使在当前这个美国表现优异的时期，我认为回顾历史也很重要。

And if we take even this good period now where US outperformed, I think it is good to remember the history.

我有一些很好的图表，展示了相对价格的变化。

And I got some nice pictures of showing the relative pricing.

美国CAPE与非美国CAPE的对比。

So US cape versus non US cape.

以及与未来十年历史的对比。

And that versus next ten years history.

它在八九十年代和二月时一直是很好的预测指标，但在全球金融危机之后的数据中却失效了。

It has been really good predictor in eighties, nineties, February, just not in the post GFC data.

所以，这一点又被忽略了。

So again, it is missing that one.

但当我们回顾这段历史时，会发现如果当时处于日经泡沫时期，1990年美国的估值仅为世界其他地区的半数。

But when we look at that history, it does tell that US actually had half the valuation of the rest of the world in 1990 if you are in the Nikkei bubble.

然后到了二月，估值水平趋于一致，达到了相似水平。

And then we got to sort of parity valuation, similar valuation levels in February.

而最近，美国的估值几乎达到了其他地区的两倍，这几乎是对日经泡沫故事的镜像。

And more recently, US got to almost double valuations, which is sort of pretty much a mirror of the Nick K.

泡沫故事。

Bubble story.

因此，历史表明，从这些水平实现超越将非常困难，因为这类估值差距隐含了远超预期1%的极端增长或持续增长优势。

So the history says it is going to be very difficult to outperform from those levels because those types of valuation gaps embed some really extreme growth or persistent growth edge much more than that 1% that's expected.

因此，我认为人们过于关注近期的证据，并过于轻率地进行了外推。

And so I do think that people have focused too much on that recent evidence and extrapolated much too lightly.

在任何时点，总会有一些故事，我这里只谈统计，但还有关于Mag7、科技优势以及如今AI等其他故事，这些都支持了这些观点。

Always at any point in time, there will be stories, and I just talk about statistics here, but there are other stories of the Mag7 and Tech Edge and now AI and so on, which support the stories.

所以这也没错。

And so that's fair.

但也许我可以这样来缓和一下我的回答：当我谈到我与后视镜预期的斗争时，我觉得我已经很好地总结了这一点。

But maybe I'll mood this answer by saying, think I summarized it decently when I was talking about my battle regarding or against rear view mirror expectations.

我说，当过度外推者最近押对了方向并因此变得过度自信时，这种后视镜心态会更加明显。

I say that rear view mirror mindset is more pronounced when over extrapolators recently got it right and become overconfident because of that.

当我们讨论股票而非债券时，这种心态也更加明显。

Also more pronounced when we look talk of stocks rather than bonds.

当过去的收益主要来自估值上升而非更可持续的因素时，这种误导性更强，而我们现在正处在历史极端水平，均值回归的风险更大。

It's more misleading when past performance came mainly from valuation increase as opposed to something more sustainable, and we are now at historical extremes where mean reversion risk is greater.

当后视镜所反映的时间跨度不够长时，比如仅以去年为基准，动量策略可能有效，但这种情况更适用于过去三到十年的区间，此时它们会更加脆弱。

And they are more vulnerable when the rear view mirror is not very long run where it might be a good base case or just last year where momentum might work, but it's something like the last three to ten years.

这是一个非常危险的后视镜视角。

That's a really dangerous rearview mirror.

而最后一个危险因素是，当出现催化剂，时代正在发生变化时。

And then the last dangerous thing is when there's a catalyst to when the times are changing.

我认为，所有这些特征都相当契合我们目前在美国和非美国市场所看到的情景，但我仍对接下来几个月做出任何明确判断持谨慎态度。

And I think all of these features somehow fit pretty well to the story we have now in The US, non US story that I still am cautious in making any calls for the next few months.

它看起来一度不错，但短期内的因素仍会主导市场走势。

It looked good for a while, but again, short term things will dominate for a while.

但如果我从三到五年的视角来看，我非常有信心：现在并不是超配美国资产的时机。

But if I take a three to five year perspective, I feel exceptionally confident that this is not the time to overweight US.

在这一系列论文的后期一篇中，你提出所需资产收益率可能已达到潜在的自然下限，我认为这是一个非常有趣的论断。

In one of the latter papers of the series so far, you argued that required asset yields have reached potentially a natural lower bound and I thought this was a really interesting claim.

因此，我希望你能详细阐述一下这一论断的依据，以及如果这一观点成立，它将如何影响投资者对未来收益预期的思考方式。

And so I was hoping you could sort of unpack what drives this claim and how, if that is true, how that impacts how investors should think about return expectations going forward.

嗯。

Yeah.

嗯。

Yeah.

有一个关于实际收益率的数百年图景，这是实际债券收益率的下降趋势。

There's a multi century picture of real yields and this is real bond yields coming down.

我先说这一点，然后我会提到另一个关于股权风险溢价的故事。

I'll first tell that, and then I say that there's actually another story which is about equity premium.

但几年前，保罗·斯梅尔钦的一篇论文，后来与肯·罗格夫和芭芭拉·罗西等人在后续研究中结合，他们基本上表明，从中世纪开始，实际债券收益率高达10%或更高，此后我们看到了一个稳步趋向零的趋势，近年来甚至出现了负的实际收益率，就像几年前我们所看到的那样。

But there was a paper, Paul Smelching's paper, a few years ago, and then it was combined with Ken Rogoff and Barbara Rossi, I think, in a later study, where they basically showed that starting from middle ages when real bond yields were like 10% or more, we got this nice trend towards zero and recent even to negative real yields as we saw now a few years ago.

这一趋势似乎暗示着，也许我们可能会跌至-5%甚至更低。

And that trend sort of suggested, hey, maybe we can go to minus 5% or something.

但在新论文中，作者指出，当我们谈论数百年的时间跨度时，这种趋势非常宽松，但零水平附近可能正在发生某些变化，使得我们未来不太可能达到-5%或-10%。

And in the new paper, the authors did say, well, you know, it's a very loose trend when we talk about hundreds of years, but also there might be something happening at the zero level, which makes it less likely we are going to hit minus five and minus 10 down the road.

我认为，从名义收益率的角度来直观理解零利率下限是有道理的，几年前我们确实看到，这个下限并没有我们原先认为的那么严格，它其实是有弹性的，但名义收益率实际上并没有跌破-1%太多。

And I think it's maybe intuitively it's good to think about with nominal yields, the zero lower bound, which again, we did see a few years ago that it wasn't quite so tight as we thought it was porous, but nominal years didn't really fall more than to minus 1% or so on.

如果通胀预期不是太高，很难想象实际收益率会低于负二到负三百分比。

And then if inflation expectations are not too high, it's hard to see much more than minus two, minus 3% real years.

因此，如果我们回到几年前的实际收益率水平，就更难论证收益率仍存在下行趋势，而更可能是均值回归和上升倾向。

So if we are getting to the kind of real yields where we were a couple of years ago, it's much harder to argue that there is still a trend downwards than mean reverse and tendency upwards.

这就是关于实际债券收益率的自然下限理论。

So that was the natural lower bound story when it comes to real bond yields.

从那以后，我们当然经历了这一转变。

And since then, we have had, of course, this shift.

十年期通胀保值债券（TIPS）收益率已从负一上升至正二百分比。

Ten year tip yield has gone from minus one to plus 2%.

顺便提一下，我一个很在意的问题是：为什么这么多所谓的长期实际资产，对十年期TIPS收益率从负一升至正二的变化无动于衷？

And just as an aside, one of my pet peeves is this question that how come so many supposedly real long term assets don't care when ten year TPS yield goes from minus one to plus 2%?

为什么房地产、公开股权、私募股权在如此重要的折现率变动下，都没有重新定价呢？

How come real estate, public equity, private equity hasn't been repriced in even such a change in the relevant part of their discount rate anyway?

但另一个体现这种自然下限的例子，当然是股权风险溢价。

But another example of this natural lower bound is, of course, then equity bond premium.

我认为人们本能地认为这一数值应该始终为正，而在20世纪的大部分时间里，它确实呈下降趋势，并在1999年、2000年左右达到历史最低点，因为当时CAPE估值处于最高水平，而实际国债收益率大约为4%左右。

And I think that there is some intuition that this should always be positive, and it certainly was coming down in much of the 1900s And with many metrics, it sort of got to all time lows around 99, 2,000, because you had the highest CAPE then, and actually treasury yields were like 4% real or something like that.

因此出现了负的股权债券溢价。

So negative equity bond premium.

但这种状况并未持续太久，此后我们看到了更高的数值。

And that didn't last long, and since then we got some higher numbers.

但从2022年起，我们再次看到股权债券溢价的收窄。

But since '22, we were getting again this compression of equity bond premium.

基本上，债券收益率上升，而股票估值几乎没有变化。

Basically bond yields rose and equity valuations hardly changed.

因此，我们虽然没有回到1999年、2000年那样的状况，但从历史角度看，当前的股权债券溢价水平确实处于极低位置。

So we are, again, we are not in '99, 2000 type of situation, but we are, from historical perspective, in particularly low level of equity bond premium.

我认为，从当前这些水平来看，很难再回到负值区间。

And I do think that that from these levels, it's really hard to go to these negative levels.

尽管我也能看到市场上存在一种极为乐观的狂热情绪，这种情形或许会短暂出现，但我认为，如果未来一两年内出现这种市场状况，它将不可持续。

Although I can also see a very bullish euphoric market out there where it could happen for a while, but I do think it's not gonna be a sustainable situation if we get that type of market in the next year or two.

我觉得我必须问个跟进的问题。

I feel like I have to ask the follow-up question.

你刚才说得有点悬而未决。

You left it hanging there a little.

我想知道，你对为什么某些资产在实际利率上升300个基点的情况下没有重新定价，有什么基本面的看法吗？

I'm curious, do you have any fundamental views as to why certain assets haven't repriced with a 300 bp increase in real rates?

这仅仅是过度热情、过度乐观，还是背后真的有基本面原因？

Is it just again over enthusiasm, over optimism, or is there actually a fundamental story there?

我认为，对于私募和公募，答案是不同的。

There's a different answer, I think, for privates and publics.

对于私募来说，它们真的不想重新定价，到目前为止也一直能躲过去。

For privates, they really don't want to reprice, and they've been able to get away with it so far.

由于市场相对冻结，交易量很少，因此存在一些不切实际的估值。

And basically with relatively frozen markets, there's not much volume where there are some unrealistic valuations out there.

这就是私募方面的情况。

That's the private side.

但在公开市场方面，我认为人工智能故事在这里起到了救世主的作用，而且这确实是基本面驱动的。

But on public side, I do think that the AI story has been a savior here, and clearly it's fundamental.

至于这是否处于泡沫阶段，这又是一个大问题。

And whether it's in bubble territory, that's, again, it's a big question.

我倾向于说，我不确定我们是在第五局还是第九局，或者用足球来比喻，是在中场休息还是加时赛，没人能真正知道。

Like, I sort of tend to say that I don't know whether we are in the fifth inning or ninth inning on that one or soccer wise halftime or overtime on that one, and nobody nobody really can know that.

我们会看到的，但如果要我猜的话，我觉得我们更接近加时赛阶段。

We'll see, but if I'd have to guess, I think we'd be closer to the overtime part.

这不仅源于对估值等数据的观察，更源于年轻投资者对风险的热衷行为——明显体现在MIM股票上，还有无杠杆的比特币、杠杆单只股票，零零零。

That comes not just looking at things like valuations and so on, but basically this the young young investors risk loving behavior with obviously MIM stocks, but zero levered this and that, levered bitcoin, levered single stocks, zero zero.

无论如何，有太多迹象表明，人们对风险缺乏敬畏。

Anyway, there's so many things that somehow reveal that there is not much respect for risk.

这在短期内会有所帮助，但最终我认为这正在埋下一些隐患，我们将在未来几年看到这些问题。

And again, that's going to help short term but eventually I think it's sowing some problems that we're gonna be seeing in the coming years.

我认为，对你系列文章持怀疑态度的读者会看到你提供的某些数据，并说：这个时间序列追溯到了一百五十年前。

I think a skeptical reader of your series would look at some of the data you provide and say, this time series goes back a hundred and fifty years.

为什么这与今天形成理性的前瞻性分析有关？

Why is that even relevant to forming a rational forward looking analysis today?

市场保持在同一地理区域，这到底意味着什么？

What does a market even, you know, staying in the same geography?

一百五十年前的美国市场，当时还是一个完全不同的构成、完全不同的风险特征、完全不同的资本市场准入方式的新兴市场，那又算什么？

What is The US market a hundred and fifty years ago when it was an emerging market with a totally different composition, totally different risk profile, totally different means of accessing capital markets?

这如何帮助我今天的决策？

How does that help inform my decision making today?

所以我想问你，你如何证明使用数百年数据来指导我们为未来十年设定的预期是合理的？

And so my question to you is how do you justify using sort of centuries of data to inform these expectations that we're trying to create for the next decade?

是的。

Yeah.

是的。

Yeah.

所以我认为，从宏观角度看是一种优势。

So I would say zooming out is a virtue.

在这系列之前，我之前的研究所涉及的是与埃洛伊·迪姆森和另外两位合作者一起研究大约一个世纪以前的数据。

And before this series, my previous study was actually with Elroy Dimson and two other coauthors studying pretty much data before a century ago.

所以如果我们能追溯到19世纪，观察资产类别的行为等等。

So really going back to 1800s if we could then look at asset class behavior and so on.

你可能看过爱德华·麦克奎里的研究结果，显示19世纪的股票与债券溢价接近于零。

And you may have seen the result by Edward Macquarie that actually 1800s equity bond premium was near zero.

我认为，这很有趣，当然会引发一些后续问题。

And I think, well, that's interesting and certainly raises follow-up questions.

但在这些后续问题中，有一些隐含在你质疑的观点里：是的，我们知道世界已经改变了。

But among the follow-up questions are the ones that were implicit in your skeptical point that, yes, we know world has changed.

我们知道数据质量更差。

We know data quality is worse.

我们是否应该完全忽视这些东西？

Should we care at all about this stuff?

我认为我们不应该完全忽略它。

I do think that we should not totally ignore it.

我认为给更近期的数据赋予更大权重是正确的，但如今的投资者往往只关注自己亲身经历的那一次体验，或许再延伸到全球金融危机之类的事情，这实在太典型了。

I think it's right to give more weight to more recent data, but it's just so typical for today's investors to only care about that one experience where their own live memories are and maybe a little longer to GFC or something like that.

但我确实认为，那些更长期的历史中蕴含着一些信息。

But I do think there is some information in those longer histories.

因此，当我们谈论CMAs或类似话题，讨论的是五到十年的预期回报——即长期视角时，仅仅研究一个十年期，简直就是笑话，或者说是误解。

So especially when we are talking about CMAs or something where we are talking about five to ten year expected returns, so long horizon, then studying just one ten year period, it's a joke or it's a misunderstanding.

你实际上只是在观察一个数据点，有时甚至只是一半的数据点。

It is really one observation that you are looking at or sometimes it's half an observation that you are looking at.

所以，即使我们回顾上个世纪，也能得到大约十个独立的观测样本。

So even when we look at last century, that would give us 10 independent observations.

事实上，如果你使用月度数据，获得一些重叠的额外数据，也不会带来太多额外信息。

The fact that if you go to monthly data, get some overlapping extra it doesn't give you too much more.

我曾在某处提到过，它可能只会让你的信息量增加到1到1.4倍左右，但绝不会带来十倍甚至百倍的信息增量。

I I mentioned it somewhere there that it might give you from one to 1.4 or something like that, but it really doesn't give you 10 times or hundreds times more information.

对于这类关于十八世纪或更早时期的问题，也许另一个有趣的回答是：尽管制度和环境发生了变化，但人类的行为并未改变。

And maybe another cute answer to this type of question on eighteen hundred's or so, yes, the institutions have changed, environment has changed, but human behavior hasn't changed.

因此，从那些过去的时代中，我们可以获得一些可能永恒不变的洞见。

So there is something interesting that we can get from those past times, which might be timeless.

总的来说，我认为更关注近期数据是正确的，但忽视过去三十年左右的数据，仅仅因为世界发生了巨大变化，我认为这是个错误的做法。

So overall, I do think that it's right to focus more on more recent data, but ignoring things beyond last thirty years or so, or because the world was so different, I think that's a bad idea.

我想在接近尾声时，将这一点与实际应用联系起来。

I wanna tie this into the practical as we get towards the end here.

从纯粹的研究兴趣出发提出资本市场假设是一回事。

And it's one thing to come up with capital market assumptions from a purely research interest.

而真正实施这些假设则是另一回事。

It's another thing to actually implement them.

因此，从你的角度来看，当投资者使用长期资本市场假设——无论是五年还是十年——并用它们来设定战略投资组合时。

And so from your perspective, when investors use long term capital market assumptions, call it five years, call it ten years, and they're using them to set the strategic portfolios.

他们应该多频繁地根据这些资本市场假设重新优化或重建投资组合？

How often should they realistically be re optimizing those portfolios or rebuilding those portfolios as informed by the capital market assumptions?

而这个问题的提出，至少在我看来，是因为每年的假设之间存在极大的重叠，对吧？

And here the question comes from, at least in my view, the fact that there's such a large overlap in the assumptions from year to year, right?

如果我们使用十年的预测并每年再平衡，那么会有九年的重叠，而且这些预测的置信区间相当宽。

If we're using ten year forecasts and we're rebalancing once a year, it's nine years of overlap and you have these fairly wide forecast confidence intervals.

因此，对我来说，问题在于如何避免在每次更新时过度拟合噪声，实际的节奏应该是怎样的？

And so to me the question becomes, what is the practical cadence here of not overfitting noise from one update to the next?

是的。

Yeah.

是的。

Yeah.

我认为至少应该每年跟踪这些数据，但具体怎么做，还要看情况——如果你是一个逆向投资者，想利用这些信息，那么你可能每六个月或十二个月就会采取一些行动。

I think it makes sense to track these things, certainly at least on an annual basis, but to make some well, again, depends if you're a contrarian investor and you wanna try to use this, then you may want to take actionable measures then even every six months or twelve months and so on.

但我认为，对许多人来说，每隔几年做一次就足够了。

But I think for many, it's fine to do this only every few years.

因为这些预测更适合作为规划工具，而不是市场择时工具。

Because these are meant to be more like planning tools as opposed to market timing tools.

如果你想进行市场择时，我认为你需要引入一些更短期的指标，而不仅仅是资本市场的假设中包含的内容。

If you want to do market timing, I think you want to bring in some more short term indicators beyond the things that are in capital market assumptions.

而且，说过去十年表现糟糕，这公平吗？当你看更长的历史数据时，预测误差或前瞻性置信区间确实相当宽泛。

And then to your, like, it's fair to say that, again, last ten years sucked, and when you look at longer histories, there are quite wide or large forecast errors or in a forward looking sense, confidence intervals.

因此，我试图在谦逊与现实之间取得平衡——这与事实相关：确实信息有限，但我也想反对一种虚无主义心态，即‘根本无法预测’，于是你可以完全忽略起始收益率，甚至说反正可以用后视镜，为什么不呢？

So I am trying to balance the humility and realism, which is related to the fact that, yeah, there's not too much information, but I also want go against the nihilistic mindset that, hey, there is no predictability, and then you could just as well ignore starting yields, you could say that you can use the rearview mirror because why not?

而我想说的是，数据并不支持这种虚无主义。

And I'm saying data doesn't support that type of nihilism.

数据显示，起始年份和估值中确实包含一些有用的信息。

Data says there is some useful information in those starting years and valuations.

让我们至少把这些信息当作规划工具来使用，并据此考虑采取一些反向策略。

Let's use that at least as a planning tool and may consider having some contrarian fields then based on them.

这些策略本质上是缓慢变化的，属于一到三年的时间尺度，真正对投资组合产生影响时才显现出来。

And they are inherently then slower moving and one to three year type of things where they really start to matter in the portfolio.

AQR自身是如何应对这些挑战的？在构建你们发布的资本市场预期时，如何平衡估值锚点、随时间发生的结构性变化，以及在形成这些指标时可能遇到的行为偏差？我记得你们是每年发布一次，你们如何避免你们在论文中提到的这些陷阱？

How does AQR itself tackle these challenges, balancing valuation anchors, structural changes that can occur over time, the behavioral pitfalls in coming up with these measures when forming the capital market assumptions that you publish, and I believe you publish them annually, how do you try to avoid all these traps that you've highlighted in the papers?

是的。

Yeah.

因此，作为起点，我将重点关注五年到十年的预测，因为基于收益率的估计在此期间尤为重要。

So as a starting point, I'm gonna focus on five to ten year estimates where yield based estimates are especially important.

当你延伸到更长的时间跨度时，初始收益率的重要性就大大降低，那时你可能只需要考虑一些合理的直觉，比如大多数资产类别的夏普比率约为0.3，再乘以相应的波动率，或者采用更基于理论的方法。

When you go to even longer horizons, those starting yields even don't matter much, and then you might want to just think of some decent intuition like 0.3 Sharpe ratio for most asset classes times the relevant volatility or something more theory based.

但如果我们能采用这种基于收益率的方法——我们尽可能这么做——顺便说一下，对于对冲基金或商品等资产，这种方法可能不适用，但在大多数资产类别中我们是可以使用的。

But if you do this yield based approach, which we do whenever we can, and by the way, we cannot maybe with hedge funds or commodities or so, it doesn't make sense, but in most asset classes we can do that.

因此，我们会说，好吧，让我们使用类似贴现现金流模型的方法，即收益率加增长，用于股权类投资。

So then we say that, Okay, let's use something like discounted cash flow model, yield plus growth for equity type of investments.

我们会认真对待这些输入参数。

And we try to be thoughtful about those inputs.

我们会引入诸如股票的净股票回购收益率和债券的滚动收益等要素。

We try to bring in things like net buyback yields and so equities and roll down effects for bonds.

因此，注重细致考量。

So thoughtfulness.

另一个我们需要认真对待的方面是增长假设，我认为你不应该让增长在时间上或国家间波动太大，因为其可预测性非常有限。

Another thing where we want to be thoughtful is with that growth, where I think you don't want to let that vary too much over time or across countries because there is so limited forecastability.

我认为还是有一些可预测性的，你希望捕捉到这一点，但如果你的估计偏离锚定值超过正负1%，我认为你可能过于随意或激进了。

I think there's some forecastability and you want to try to capture that, but if you do more than plusminus 1% around an anchor, I think you are probably too overconvenient or aggressive.

我们在这方面缺乏足够的信息。

We just don't have enough information on that.

因此，我对这些增长预测进行了相当程度的锚定。

So I'm sort of anchoring those growth estimates quite a bit.

当然，我们还会排除分析师的增长预测——尽管从逻辑上讲，使用这些预测似乎很有道理，但证据明确显示，他们的过度外推会使这些预测成为未来增长或未来市场回报的反向预测指标，因此你不应该纳入它们。

And certainly excluding, by the way, those analysts' growth forecasts, which again logically would make a lot of sense to use them, but because the evidence is so clear that their overextrapolation makes them inverse predictors of future growth or future market returns, you don't want to include.

此外，还要进行模型平均。

Then in addition to this, do model averaging.

我们不知道哪个模型是正确的，所以不妨结合几种不同的输入方式。

We don't know what's the right model, so let's combine a couple of different types of inputs there.

这是一点。

That's something.

尤其是在均值回归问题上。

And then especially on the mean reversal question.

因此，历史数据表明，实际债券收益率、股票估值，以及可能的利润率增长等方面，似乎存在某种均值回归现象，我们可以找到一些表现出长期均值回归特征的序列。

So historical data would suggest for real bond yields, there seems to be some mean reversion, equity valuations, and maybe margin growth or something like we could find some series which have exhibited some long run mean reversion.

但它们也表现出相当持久的趋势，尤其是在最近几十年。

But they also have shown quite persistent trends, and especially in recent decades.

因此，我们某种程度上选择不包含这一部分，即可能存在结构性变化，而我们已经接收到的信息是：高估值会带来低起始收益率，这足以让我们保持逆向思维，而无需再叠加均值回归，否则可能会使效果加倍。

And so somehow we've chosen not to include that part there saying that there may have been a structural change, and we are already getting the message of, let's say, high valuations will give us low starting yields, and that's sort of enough for us to get the contrarian spirit without adding the mean reversion, which would, I don't know, double up the effect there.

因此，我们允许结构性变化的可能性削弱均值回归的逻辑，而这确实是一个极具争议的问题。

And so we are allowing this possibility of structural shifts to take away the mean reversion logic, and that's a really debatable thing.

如果没有沟通简洁性的考量，我可能会假设20%的均值回归幅度。

If there wasn't a communication simplicity argument, I would probably assume 20% mean reversion.

因此，在五到十年的时间范围内，存在温和的均值回归，大致如此。

So, mild mean reversion in over a five to ten year horizon, something like that.

这将是基准情景。

That would be the base case.

我很庆幸我们没有那样做，因为过去十年里，那样做会使我们更加看空。

I'm glad we didn't do that because last ten years that would have made us even more bearish.

我一个蹩脚的笑话是，当人们调侃我们对美国股市回报过于谨慎时，至少我们没有遭遇糟糕的一年，也没有采用GMO的均值回归假设。

And one of my lame jokes is when people tease us about our cautious US equity returns, well, at least we didn't have the rough year and GMO mean reversion assumption there.

发布研究者和数据的一个风险是，人们可能会以你未曾预料的方式使用它。

One of the risks of publishing researcher and data in particular is that people might use it in a way you didn't intend.

对吧？

Right?

因此，你开发了这些资本市场假设，并将它们公之于众。

So you develop these capital market assumptions and you publish them to the world.

如果你是负责配置资产的决策者，正在消化这些资本市场假设，你希望它们在实际配置中被赋予多大的权重？

If you were in the driver's seat for an allocator who is ingesting those capital market assumptions, what's your preferred weight in which they'd be deployed to inform their actual allocations?

我认为它们应该被使用，但要保持谦逊。

I think they should be somehow used, but with humility.

因此，我们再次设置了置信区间，以强调这些假设可能严重偏离现实。

So again, we do put some confidence bands there to highlight that these things can be quite wrong.

因此，基于它们采取过于激进的行动，我认为总是不明智的，但可以适度倾向它们，或许用它们来帮助我们在类似当前美国与非美国市场这样的环境中保持耐心。

So very aggressive actions based on them are, I think, always a bad idea, but tilting towards them and maybe using them to try to remain patient when we are in a situation like we have been, for example, with US, non US.

如果这些观点能帮助人们坚持一种现在看来比几年前更有道理的立场，那就有帮助了。

If these things help people stick with a view which probably makes even more sense now than a few years ago, then that's helpful.

但任何非常激进的观点，我认为我都应该保持谨慎。

But any very aggressive views, I think I'd be quite cautious.

这里的预测能力还不够。

And this just isn't enough predictive ability here.

我们几个月前就根据你非常繁忙的日程和频繁的出差安排了这次采访。

So we scheduled this interview, I think months ago around your very busy schedule, very busy travel schedule.

但不幸的是，这个系列还没结束。

And unfortunately, the series isn't over.

我本想在系列结束时再做这次采访，以便获得完整的视角。

I would have loved to have done this at the end of the series and gotten the full picture.

但先给我一个预告吧。

But give me a preview.

接下来会发生什么？

What is to come?

在系列的剩余部分中，你还有什么需要探索的？

What do you still have to explore in the rest of the series?

我们和汤姆一起发布了七篇文章，我想还剩三篇要完成。

So we published seven pieces with Tom, and we've got, I think, three to go.

可能还有第四篇。

Might be a fourth one.

我们拭目以待。

We'll see.

基本上，会有两篇关于债券市场的文章。

So basically, there will be two on bond markets.

其中一篇将深入探讨收益率曲线的故事，即均值回归的利率预期。

One will drill into this yield curve story, mean reverting rate expectations.

我认为这不仅仅是重申我们之前文章中讲过的简单要点。

Think there's something more than just to highlight sort of the elevator pitch story that we've told in an earlier piece.

但另一篇则专注于债券收益率水平。

But another one is just on bond yield levels.

这正是我几十年来一直撰写的内容。

Think this is what I've written about for decades.

这是我最喜欢思考的问题之一：用调查数据分解十年期收益率。

It's my sort of favorite thing to think about decomposing just ten year yield with survey data.

你可以将十年期收益率大致理解为通胀预期、实际短期利率预期和所需的期限溢价。

And you can basically think of ten year yield as inflation expectations, roughly real short rate expectations and required term premium.

用调查数据来做这种分析很有趣。

And it's interesting to do that with survey data.

近年来，故事基本上是：我们从极低水平看到债券收益率上升，因为实际部分——r星和期限溢价——从负值转为正值，而通胀预期几乎没有任何变化。

And in recent years, the story is basically that we got this increase in bond yields from these very low levels because the both real parts, R star and term premium went from negative to positive, while inflation expectations didn't move almost at all.

我认为这在当下极具相关性，因为显然几年前通胀预期依然保持稳定，这得益于美联储的信誉。

And I think this is highly topical now because clearly the inflation expectations stayed as well anchored a few years ago because of Fed's credibility.

而如今，政府几乎准备要玩火了。

And right now, the administration is pretty much ready to play with fire there.

因此，我认为现在确实存在一种主观判断，但若基于这类分析并加入一些主观判断，我们就有比几十年来更好的机会，避免长期通胀预期脱锚，而这将对债券市场乃至许多其他资产构成严重问题。

And so I do think that there is a sort of this is a discretionary view here now, but based all that type of analysis and adding a bit of discretion, then we have a better chance than in many decades to see long run inflation expectations getting de anchored, and that would be a pretty serious issue for not just for bond markets, but for many other assets.

不知道这是否会真的发生，但即便在我的职业生涯中，这种可能性也更大了。

Don't know whether that's going to happen, but again, chances are bigger than even in my career.

事实上，我在1970年代就已活跃于市场。

I was in fact active in 1970s.

所以这是债券部分。

And so that's a bond part.

而我们目前思考的最后一部分，基本上是后视镜和流动性分散工具。

And the very last piece that we are currently thinking is basically rearview mirror and liquid diversifiers.

所以这差不多也接近你的主场了。

So that's getting close to also your home court there.

我认为投资者显然很清楚，最近几年表现不错，但如果我们只看过去十年，包括2010年某些流动性策略（如商品、价值或趋势策略）表现糟糕的时期，这些策略对那段时期并不宽容，相比之下，我认为更长的历史周期才更相关。

I think like investors clearly, I mean, last few years have been pretty good, but let's say that you often think of just last ten years, including then the bad times in 2010 for some liquid strategies, could be commodities, be value, could be trend, and they are not so forgiving on that one compared to especially much longer histories which I think are more relevant.

因此，我认为后视镜效应抑制了投资者对流动性分散工具的采纳，而投资者显然更倾向于接受非流动性替代品。

So I think the rear view mirror has been discouraging investor adoption of liquid diversifiers, whereas investors have been obviously embracing illiquid alternatives.

在这方面，后视镜效应甚至可能引发对数据的争议，但毫无疑问，当我们思考为何私募股权等资产表现如此出色时，当时的估值区间更宽、杠杆成本更低，而这些条件如今已不复存在。

And there I think the rear view mirror, one can even debate the data, but certainly when you look at why they would have had as good performers, thinking of for example private equity, there were basically conditions of much wider valuation band, much cheaper leverage than what are available now.

因此，我认为，在当前估值水平和杠杆成本不变的情况下，私募股权很难再实现任何类型的超额收益。

So I do think that it's very difficult for, let's say, private equity now to deliver any kind of outperformance as long as we have what the valuations where they are now and cost of leverage where it is now.

我们的收费水平也是一个因素。

And our fee level is another one.

如果他们能显著降低费用，这将是解决预期回报问题的另一种方式。

If they were to lower the fees meaningfully, that would be another way of solving the expected return problem.

但我们的估算表明，私募基金历史上表现优异，但近期前景相当黯淡，尽管未来情况可能会发生变化，但这需要在费用、利率或私募估值方面做出调整。

But our estimates say that historically great performance for privates prospectively, pretty gloomy near term, and eventually things can change, but it will require changes in either fees or rate level or valuations in those privates.

所以，这些内容仍在推进中，除非我们决定再做一篇总结性的内容，这一点还有待决定。

So anyway, so those are still in the pipeline, and then probably over unless we decide to do one last concluding piece, and that's to be seen.

好吧，安蒂，趁你现在在这儿，如果我不问你一个关于股票和债券领域我视为最具开创性的论文之一的问题，那就太遗憾了，这篇论文是你在大约2003年之前写的。

Well, Antti, while I have you, I'd be remiss if I didn't ask a question about one of the papers that I consider to be one of the seminal papers ever written on stocks and bonds, which is a paper you wrote back in probably pre two thousand three.

我知道它发表于2003年，考虑到期刊的流程，你很可能在2002年就写好了，但它的标题恰如其分——《股票与债券的相关性》。

I know it was published in 2003, knowing the journal process, he probably wrote it in 2002, but it's the appropriately titled Stock Bond Correlations.

这篇论文我曾在2022年股票和债券相关性飙升时，发给了数十人，当时对他们来说这是一种不寻常的体验，他们正在寻找更根本的论据和解释。

And this is a paper that I must have sent to dozens and dozens of people in 2022 when stock bond correlations were spiking and this was an unusual experience for them and they were looking for more fundamental arguments and explanations.

我想问你关于那篇论文的问题，其实并不是特指那篇论文，而是那篇写于二十多年前的论文，其相关性在二十年后反而显著提升，我认为它在基本论点上是正确的。

What I wanted to ask you about that paper and really it's not about that paper specifically, but that's a paper that was written twenty plus years ago whose relevancy really heightened twenty years later and I think proved correct in its fundamental arguments.

但我的问题是，如果你今天重写这篇论文，有没有什么从中学到的经验、想法的改变，或者你希望添加或澄清的新观点？

But my question is, is there anything about that paper that if you were to rewrite it today, lessons learned or maybe ideas changed or novel thinking that you would like to add or clarify to that paper?

是的。

Yeah.

我们几年前，大概是2022或2023年，和几位同事一起写过另一篇论文，乔丹·布鲁克斯早在2010年就有了很好的想法，我们后来把这些都整合在一起了。

We did write another paper just a few years, maybe '22, '23, then with several colleagues like Jordan Brooks had already a nice idea in 2010, and then we put all of that together.

这篇论文可能是JOPM论文，布里克斯顿和我们另外五到六个人共同完成的。

This is maybe JOPM paper, Brixton and all, five or six of us in that paper.

在原始论文中，我们提出了许多关于股票债券相关性可能反转的原因。

So in the original paper, there were many points why the Stockholm correlation might flip sign.

顺便说一句，我对自己在1998年LTCM危机前后就预感到这种相关性可能从正转负，感到相当自豪。

By the way, I do feel pretty proud that I was sort of seeing it happening maybe like 98 LTCM time, was starting to think, is this going to turn from positive to negative?

我觉得我当时对这一点的判断相当准确，之后我更贪婪地思考，我们是否能预测另一部分重新转正。

And I think I sort of got that pretty well and then greedily have been thinking whether we can get the other part flipping positive.

如果我们必须从我之前强调的几件事中挑选一个故事，那就是水平依赖的通胀不确定性，也就是说，在更高的通胀水平下，股票和债券的相关性会更强。

If we'd have to pick one story out of the few things that I was highlighting there, it was sort of level dependent inflation uncertainty, which could be just then said that at higher inflation levels, there's more stock bond correlation.

但我认为，乔丹的简化模型非常有用，它让我们思考股票和债券实际上是受增长和通胀因素驱动的。

But I think this Jordan's toy model turned out to be really useful to say that let's just think about stocks and bonds being driven by growth and inflation use.

如果我们面临与增长相关的不确定性，这种不确定性通常会使两种资产走向相反方向，那么就会出现负的股票-债券相关性。

And if we have got growth related uncertainty, which tends to push two assets to opposite directions, then we have a negative stock bond correlation.

如果我们面临更多的通胀相关不确定性，这种不确定性通常会使两种资产朝同一方向移动，那么股票-债券相关性就会变得更正。

If you have more inflation related uncertainty, which tends to push both assets to the same direction, we have got more positive stock bond correlation.

这个故事很好地解释了为什么在20世纪末，股票和债券呈现正相关性。

Well, that storyline tells very nicely why in late 1900s we had a positive stock bond correlation.

而一旦通胀预期在2000年左右稳定下来，只剩下增长不确定性，于是出现了长达二十年的负股票-债券相关性，因为增长不确定性占主导地位。

And once inflation expectations stabilized around 2000, there was only growth uncertainty and twenty years of negative stock bond correlation because growth uncertainty dominated.

因此，根据这个故事，过去几年里，股票与债券相关性的符号主要取决于我们是否出现了通胀预期的脱锚。

And so with that story, it's been the case now for the last few years that the sign of stock bond correlation is going to be primarily driven by, do we get this de anchoring of inflation expectations?

我们在2022年感受到了这种迹象，但并不强烈，因此目前我们大致徘徊在零附近。

And we get a sense of that in 'twenty two, but not so much, so we are sort of hovering nowadays near that zero.

因此，我非常有信心地说，股票与债券相关性的未来主要取决于通胀预期的变化。

Again, I feel very confident then in saying that the future of the stock board correlation is primarily going to be driven by what happens to those inflation expectations.

并不是通胀水平本身，而是与该水平相关的通胀不确定性。

Not really the level, but related to that level comes the inflation uncertainty.

我认为，这种不确定性将推动股票与债券相关性转向正值。

And this uncertainty would, I think, push stock board correlation then to positive levels.

如果市场在自满中是正确的，我们只是进入一个2%的平淡、稳定的通胀世界，那么未来我们可能仍会维持负的股票与债券相关性。

If markets in their complacency are right and we just go to 2% humdrum type of inflation, stable inflation world, then we probably will have still negative stock board correlation in the future.

因此，我对第二篇论文非常满意，它澄清了这一机制——不仅仅是通胀水平，而是通胀不确定性与增长不确定性之间的区别。

So, think I'm quite happy with that second paper clarifying the mechanism and that it's not just inflation level, but it's really inflation uncertainty versus growth uncertainty.

此外，我们还探讨了其他几个方面。

Plus, we do cover several other things.

增长与通胀是正相关还是负相关，这一点很重要。

It matters whether growth and inflation are positively or negatively correlated.

你可以思考信用风险的作用。

You can think about the role of credit risk.

所有这些都假设我们讨论的是无风险的政府债券，但如果我们谈论的是希腊或某些新兴市场，它们存在很高的信用风险，情况就大不相同了。

All of this sort of presumes that we are talking really of riskless government bonds, whereas if we are talking of Greece sometime or some emerging markets where they have got lots of credit risk, it's a very different story.

现在情况变得有趣了，因为即使在美国，政府赤字也引发了一些警报，这可能是另一个推高相关性的逻辑。

That's getting interesting now when the government deficits even in US have raised some alarm bells, and that could be another logic that could push the correlation higher.

但无论如何，这些都只是次要故事，主要故事还是关于通胀不确定性。

But anyway, those are sort of the side stories, the big stories about those inflation uncertainty.

好的，安蒂。

Alright, Antti.

我们今天这期节目到这里就结束了，我想问你一个我这个季节问每位嘉宾的问题：工作中之外，有什么东西目前让你着迷？

We have come to the end of the episode here, and I wanna ask you the question that I ask every guest for this season, which is what is something outside of work that you are currently obsessed with?

它可以是一个想法、一种经历、一项爱好，任何真正吸引你注意力的事情，我知道你的第一反应会是‘预期收益’。

It could be an idea, an experience, a hobby, something that has really captured your attention and I know your default would be to say expected returns.

你不能说预期收益，但我要的是工作之外、让你着迷的东西。

You cannot say expected returns, but something outside the realm of work that has you obsessed.

我一直很关注工作与生活的平衡，几十年来，我在家庭、朋友以及关注这些事情上都做得不错，因为这确实值得。

I have cared about balancing work and life and I'm I'm I'm pretty happy with how I've done it with family and with friends and paying attention to such things for decades because it does pay off.

这真的像是一个60岁老人说的话，考虑到你和年轻人的处境，你最好多关注他们，但也要试着保持其他方面的意识。

It's really like this comes from a 60 year old guy, also with your situation with the youngs, well, you better focus on those, but try to keep some other things in mind.

现在具体来说，我来自芬兰，一直保持着与那里的联系。

And now specifically then, I'm originally from Finland and I've always had my connection.

我有很多那时的朋友，我不确定。

I've got lots of friends from that time, I I don't know.

故乡的召唤对我越来越强烈。

The roots are calling stronger for me.

我想花更多时间在那里，部分是因为我热爱那个国家、自然风光等等，但那里也有我许多家人和朋友，这真是太棒了，我也建议其他人多考虑这种温情的想法。

I wanna spend more time there partly because I love the country, the nature and so on, but I also have lots of family and friends there, which is just wonderful and so recommending playing to such softy types of thoughts also for others.

潘蒂，总是很愉快。

Panti, always a pleasure.

非常感谢你加入我。

Thank you so much for joining me.

这太棒了。

This was great.