Markets Weekly August 16, 2025

Hello my friends, today is August 16th and this is markets weekly. So this past week was kind of a slow summer weekend market again where we just kind of had a modest melt up. However, we did have some interesting data. So today let's talk about three things. First off, let's talk about the interesting idea that the rise in unemployment rate among new grads is due to replacement from AI. Let's take a look at this from two perspectives. Secondly, this past week was all about inflation. We had CPI, we had PPI and we had import prices. Let's take a look at the data, see how the market received it and see how this might impact the Fed. And lastly, now I'm getting the sense that we are in some kind of dot com boom, particularly when it comes to AI. This is a mean of can't go higher, but of course, let's also look at how this compares to the dot com boom among some metrics.

你好，我的朋友们，今天是8月16日，这里是《市场周报》。上周整体是一个缓慢的夏日市场行情，市场略有上涨。不过，我们也获得了一些有趣的数据。所以今天我们来聊三件事。 首先，我们讨论一下一个有趣的观点，即新毕业生失业率的上升是由于人工智能的替代。我们从两个角度来看这个问题。 其次，上周的焦点是通货膨胀。我们有消费者物价指数（CPI）、生产者物价指数（PPI）和进口价格的数据。我们看看这些数据，看看市场的反应，以及这可能对美联储的影响。 最后，我感觉我们正处于某种类似互联网泡沫的繁荣中，尤其是在人工智能领域。这并不意味着无法达到更高的高度，但我们也要看看在一些指标上这与互联网泡沫的比较。

All right, starting with the labor market. So as we've all observed, it's hard and harder for new college grads to find work. Now one of the leading theories for this is due to AI. Now according to this interesting research piece from the Burning Glass Institute, AI could be replacing young workers. Now first off, let's take a step back and look at what AI is good at. AI is good at customer service, writing, editing, basically data stuff and stuff that you would expect a junior employee to do. For example, if you were a new investment banker, you would be building models, you would be making pitch books.

好的，先谈谈劳动力市场。大家都观察到，现在新毕业的大学生找工作越来越困难。一个主要的理论认为，这跟人工智能有关。根据Burning Glass Institute的一项有趣研究，人工智能可能正在取代年轻员工。首先，让我们退一步，看看人工智能擅长什么。人工智能擅长客户服务、写作、编辑，基本上是处理数据之类的工作，而这些工作通常是新人会做的。例如，如果你是一个新入行的投资银行家，你可能会制作模型和准备推介材料。

And today that work could be done basically instantly by AI and oftentimes much better. And we do hear anecdotal reports of firms using more AI and just hiring fewer general junior employees. Now looking at the data for the unemployment rate among people in the early 20s, segmented by degree status, we noticed that recently there's been a jump in the unemployment rate among new college grads. However, when you look at young people who don't have a college degree or who have an associate's degree, you don't really see that jump in unemployment. So this seems to be a phenomenon that's confined to new college grads.

今天，这些工作可以基本上由人工智能瞬间完成，而且效果往往更好。我们确实听到一些公司在使用更多的人工智能，并因此减少雇佣普通初级员工的例子。现在，根据我们查看的按学位状况划分的20岁出头人群的失业率数据，我们注意到最近新毕业大学生的失业率有了一个跳升。然而，当你观察没有大学学位或拥有副学士学位的年轻人时，并没有看到失业率的这种跳升。因此，这似乎是一个仅限于新毕业大学生的现象。

Taking a step further, when you look at the unemployment rate by specific college discipline, there's a notable jump in the unemployment rate among new grads in computer science. And as we've all read, AI adoption is most prevalent in the tech sector. They are, after all, most technically savvy. So all of this kind of makes sense. And then there are also other data points that suggest there's been an increase in the demand for more experienced workers and a relative decline in demand for less experienced workers, which also makes sense because if you are a more experienced worker, you can leverage your experience and basically just use AI to have an unlimited amount of junior employees. You basically become more viable and more productive.

进一步来看，当你按具体大学学科查看失业率时，会发现计算机科学应届毕业生的失业率显著增加。正如我们所读到的，人工智能技术的应用在科技行业最为普遍。毕竟，他们在技术上是最为精通的。因此，这一切都是合理的。此外，还有其他数据表明，对更有经验的员工的需求增加，而对经验较少的员工的需求相对下降。这也很合理，因为如果你是经验丰富的员工，可以利用自己的经验，基本上用AI代替无数初级员工，从而使自己更有竞争力和生产力。

So looking at the data, it all kind of, you know, it incorporates this basic thesis that firms are using AI more and AI is replacing junior employees. However, let's take a look at research done by a different team of researchers who come to a different conclusion. First off, their approach is to segment the labor market according to vulnerability to AI. According to them, some jobs are much more vulnerable to AI than others. Something that's not very vulnerable would be like a ballerina or some kind of handyman job, whereas if you are someone who's like a financial data analyst, your job is super vulnerable to AI.

根据这些数据，整体来看，它体现了一个基本的观点：即公司越来越多地使用人工智能，而人工智能正在取代初级员工。然而，让我们来看看另一个研究团队的研究结果，他们得出了不同的结论。首先，他们的方法是根据对人工智能的脆弱程度来划分劳动力市场。在他们看来，有些工作比其他工作更容易受到人工智能的影响。例如，一名芭蕾舞演员或某种手工匠的工作不太容易受到影响，而像金融数据分析师这样的职位则极其容易受到人工智能的影响。

So they categorize job vulnerability from one to five with one being the least vulnerable. Just looking at the basic demographics of these industries, you notice that overall college grads are the most vulnerable to AI, whereas blue-collar men are the least vulnerable. So if there is an impact of AI on jobs, we'd expect the unemployment rate to rise among jobs among occupations that are most vulnerable to AI. But when they break the data down, that's just not very clear. You can see that there's no notable spike in the unemployment rate among occupations that are most vulnerable to AI.

他们对工作的脆弱性进行评级，从一到五，一是最不脆弱的。根据这些行业的基本人口统计数据来看，整体上，大学毕业生对AI的脆弱性最大，而蓝领男性的脆弱性最小。所以，如果AI对就业有影响，我们会预计在那些对AI最脆弱的职业中，失业率会上升。但当他们细分数据时，这点并不明显。可以看到，在那些对AI最脆弱的职业中，失业率并没有显著上升。

But that's not the only way to look at it. It's also possible that if you are a worker that got displaced by AI, you're just straight out, unemployable, think about the boomer who's doing data analysis his entire life. He can obviously train to become a ballerina or something like that, so he just leaves there, a labor force. But when you break down the data by people leaving the labor force, it's not really clear that jobs that are most vulnerable to AI displacement are suffering from particularly higher rate of people exiting labor force.

但这并不是唯一的看法。还有一种可能性是，如果你是因为人工智能而丢掉工作的员工，你可能完全无法就业。想想那个做了一辈子数据分析的婴儿潮一代。他显然可以去学习成为一名芭蕾舞者或类似的职业，所以他就这样离开了劳动力市场。但当你分析那些离开劳动力市场的人的数据时，并不清楚那些容易被人工智能取代的工作是否有更高的退出率。

Another way to slice the data is just to look at this on a firm wide basis, our firms that are most vulnerable to AI shrinking. Maybe that's how it shows up. When you look at the employment levels of firms in the category of jobs that are most vulnerable to AI, you don't actually see that either. So far, you don't really see AI having an obvious impact. But this team of researchers also directly addressed the claim that AI is having a negative impact on new college grads. Now when you break down new college grads by the employment they're choosing, you can see that the unemployment rate for new college grads are rising across the board, not just in industries where they work that are most vulnerable to AI.

另一种分析数据的方法是从整个公司层面来看：我们的公司中哪些最容易受到人工智能影响而萎缩。这可能就是它的表现方式。当你查看那些工作类型最容易受人工智能影响的公司的就业水平时，你实际上并没有发现这种情况。到目前为止，还没有明显看到人工智能产生明显影响。但这组研究人员也直接回应了人工智能对新毕业大学生产生负面影响的说法。现在，当你按新毕业生所选择的就业类型进行分析时，你会发现新毕业生的失业率在各个行业中都在上升，而不仅限于那些最容易受人工智能影响的行业。

So their conclusion is that AI is not really having a noticeable impact on the labor market, at least not yet. And so maybe the rise in unemployment that we see among new grads is more of a macroeconomic phenomenon. Maybe it's a slowing at the economy rather than AI. So two perspectives here, I personally find the second perspective to be more persuasive. I personally definitely use AI every day. I think it's a really good way to search the web. Definitely makes my life easier. But it doesn't seem so far that it's been rising on the macroeconomic scale to be replacing new grads in any significant way. That could happen, of course. So it remains to be seen.

他们的结论是，人工智能目前并没有对劳动力市场产生明显影响，至少还没有。因此，可能我们看到的新毕业生失业率上升更多是宏观经济现象的结果，可能是经济放缓的原因，而不是人工智能导致的。关于这点有两种观点，我个人觉得第二种观点更有说服力。我每天都会使用人工智能，觉得它是一个很好的网络搜索工具，确实让我的生活变得更轻松。但到目前为止，人工智能似乎还没有在宏观经济层面上对新毕业生造成显著的替代作用。当然，这种情况有可能会发生，所以还需要继续观察。

All right, the second thing that we want to talk about is inflation. So as we all know, we are kind of in an interesting place where labor market is weakening, but inflation remains pretty subrenally high above the Fed's target. And we all want to know what the Fed is going to do. There's a debate on the FOMC right now where we had two notable dissents last meeting and maybe we'll have more dissents going forward. The market at the moment is pricing in a pretty firmly a September cut.

好的，我们要谈的第二件事情是通货膨胀。正如我们所知道的，我们现在处于一个有趣的局面：劳动力市场在减弱，但通货膨胀仍然非常高，超过了美联储的目标。我们都想知道美联储将如何应对。目前在美联储联邦公开市场委员会（FOMC）内部存在争议，上次会议有两位成员表示不同意见，未来可能会有更多不同意见。当前市场普遍预期美联储将在九月降息。

So last week we got CPI and CPI actually, I thought it had a pretty interesting market reaction. So headline CPI was a little bit lower than expected, but core CPI was a little bit higher than expected. And when you look at this from an annualized perspective, it's really clear that inflation seems to be heading higher at least based on the most recent CPI print. Or at the very least, if you zoom out a little bit, it's really clear that inflation is just kind of stuck around 3%, according to CPI, higher of course than the Fed's target. But the market took one look at that CPI print and began, the bond market began to rally fiercely and the market began to be more sure that we would have rate cuts in the coming meetings.

上周我们得到了消费者价格指数（CPI）数据，我认为市场对此的反应非常有趣。总体CPI数据略低于预期，但核心CPI数据却略高于预期。如果从年化的角度来看，很明显基于最新的CPI数据显示通胀似乎正在上升，或者至少从更长的时间跨度来看，通胀基本上停留在3%左右，显然高于美联储的目标。不过，市场在看到这一CPI数据后，债券市场开始激烈上涨，市场对未来几次会议将降息的预期变得更加坚定。

Now that's kind of interesting, but that could simply be that the market was kind of afraid or braced for a worse print and it got a better than expected print. But looking at the data itself, it's just really clear that inflation is not at least trending lower. Now one interesting thing that people have been looking at is for signs that you have some kind of inflation driven by tariffs. Now, the tariff pass who really wasn't very obvious in the last print, what was more interesting is that inflation seems to be led by increases in services. Now to be clear, some of the services inflation could be due to, you know, also indirectly due to tariffs, maybe some vendor, services vendor simply is importing some of their raw material, something like that and passing that on.

这挺有趣的，但可能只是市场原本担心或准备迎接更糟的结果，而实际数据比预期要好。不过，从数据本身来看，很明显通货膨胀的趋势至少没有下降。大家关注的一个有趣点是，是否有迹象表明通胀是由关税引起的。不过，在最新的数据中，关税影响并不明显。更有趣的是，通胀似乎主要来源于服务价格的上涨。需要说明的是，部分服务价格的上涨也可能间接受到关税的影响，比如一些服务供应商可能进口了原材料，因此将这些成本转嫁到了价格上。

But oh no, it's not super clean the data that there's been a clear pass through. Now the second set of inflation prints we got were PPI, that is prices received by US producers. Now PPI was surprisingly much higher than market expectations and it elested a pretty strong reaction from the market. The market basically unwelmed a lot of the rallying bonds that we saw from CPI, although of course it's still firmly priced in a September cut. Now PPI services was also particularly hot, but a lot of data is basically margins.

哦，不，数据不是非常清晰，显示出有明显的传导效应。我们现在得到的第二批通胀数据是PPI，也就是美国生产者收到的价格。PPI比市场预期高出不少，这引发了市场的强烈反应。尽管市场仍然坚信九月份会降息，但CPI引发的债券涨势大体上被削弱了。PPI服务数据也尤其高，不过很多数据基本上反映的是利润率。

Now when you're thinking about margins from PPI, it's important to note that it's a very volatile series. For example, let's say that you are a wholesaler that imports stuff for $98 and then sells it to someone for $100. Remargin is 2%, you know, even if that margin would go to 3%, that would be a 50% increase, but at the end of the day it's only 3%, and it's not going to have a big price impact on whoever you sell it to. So this PPI services margin is a very volatile series. And again, as we usually note, what matters to the Fed is not CPI or PPI, but ultimately what matters is PCE.

现在，当你考虑PPI（生产者价格指数）带来的利润率时，需要注意这是一个非常波动的序列。比如说，你是一名批发商，以98美元的价格进口商品，然后以100美元的价格卖给别人。利润率是2%，即使这个利润率增加到3%，增幅也达到了50%。但说到底，也只是3%，对购买者的价格影响不大。因此这个PPI服务利润率是一个非常波动的系列。再强调一下，我们通常指出，对美联储来说重要的不是CPI（消费者价格指数）或PPI，最终重要的是PCE（个人消费支出）。

Now when you have CPI and PC, you're going to make a pretty good calculation as to what PCE will be. According to the Cleveland Fed, PCE is going to be, you know, let's say a little bit below 3%, which of course is notably above the Fed's 2% target, but again, it just kind of been stuck there for some time. So looking at this data, what is the Fed going to do? Now the Fed isn't a pretty tough spot because as the labor market appears to be weakening, it seems like inflation is also stuck as well.

现在，当你手上有消费者价格指数（CPI）和个人消费（PC）数据时，你可以大致预测出个人消费支出（PCE）的情况。根据克利夫兰联邦储备银行的数据，PCE预计会略低于3%。这显然高于美联储设定的2%的目标，但它已经在这个水平上停留了一段时间。面对这些数据，美联储会怎么做呢？现在，美联储处于一个相当困难的位置，因为劳动力市场似乎正在走弱，而通货膨胀率似乎也停滞不前。

Part of that inflation could be due to tariffs, and so you have more dovish peoples like over the wall I would say that. So this tariff stuff fits a one-time increase in the price level. It's transitory after the firms make this one-time increase in prices due to tariffs. If you're not going to do the same, after all, there's not going to be additional tariffs, at least we don't expect them to be additional tariffs next year. And at the same time, science at the labor market is weakening, so we've got to cut.

部分通货膨胀可能是由于关税造成的，因此你会发现有人态度更为鸽派。我认为这种关税问题导致的物价水平上涨是一次性的。在企业因为关税而进行一次性提价后，这种影响是暂时的。如果不再增加新的关税——毕竟我们预计明年不会有额外的关税——那么这种影响就不会持续。同时，劳动力市场呈现疲软的迹象，因此我们需要采取降息措施。

And it seems like the market is firmly believing that the Fed at least will cut once September, maybe twice, again, maybe in December, but that's basically the market's expectations. A more hawkish person could say, and this is still totally possible, that we just shouldn't cut it off. Policy is a good word is. Labor market is strong. Yeah, what looks like weakness is simply that we have a reduced increase in supply of labor due to lower immigration, overall unemployment rate is low.

市场似乎坚信，美联储至少会在九月份降息一次，也可能降息两次，再可能在十二月份降息一次。但这基本上就是市场的预期。不过，持更强硬观点的人可能会说，这种预期也有可能完全不成立，我们就不应该降息。货币政策目前是合适的。劳动力市场也很强劲，看似的疲软只是由于移民减少导致的劳动力供给增长减缓，总体失业率依然很低。

Now, sure, Powell is going to speak at Jackson Honex week, and it's possible he could tilt his hand to show a little bit of how he's leaning. But with such disagreement on the Fed, my bet is that Jackson Hose is just not super and ventful, maybe talking more about the importance of Fed independence and so forth. After all, he still doesn't have the next non-farm payrolls report, which I think would be decisive when it comes to whether or not the labor market is weakening.

当然，鲍威尔将在下周的杰克逊霍尔经济政策研讨会上发言，他可能会稍微透露一下他的倾向。但由于美联储内部意见分歧，我认为杰克逊霍尔会议不会有太多重大事件，可能更多地讨论美联储独立性的重要性等话题。毕竟，他还没有得到下一份非农就业报告，而我认为这份报告对于判断劳动力市场是否走弱是决定性的。

So the moment the market is happy, firmly pressing again, cut at September, another cut by the end of the year, and beyond that, we'll just have to see where the data falls. Okay, the last thing that I want to talk about is that there are just more and more signs to me that we are in some kind of dot com style boom that inevitably will be followed by some sort of dot com bust. But to be clear, these things can go higher.

所以，一旦市场情绪乐观，再次坚定压下去，预计在九月降息一次，然后在年底前再降息一次。至于之后的情况，我们需要看数据如何表现。最后我想说的是，我越来越觉得我们正处于某种类似互联网泡沫的繁荣期，而这很可能会不可避免地导致类似的泡沫破灭。但需要明确的是，这种情况可能会继续升温。

It doesn't mean you can't make money. It doesn't mean we can't go much higher. But I think it does significantly really increase the possibility that we could also go much lower. Now, there's a lot of different ways to look at the markets when you come to this. But I think some popular metrics, again, these metrics obviously don't really matter because of traditional fundamental metrics. It didn't matter at all. We wouldn't be where we are today.

这并不意味着你不能赚钱，也不意味着我们不能涨得更高。但我认为这确实大大增加了我们可能会跌得更低的可能性。现在，分析市场有很多不同的方法。但是，我认为一些常用的指标，其实在传统的基本面分析中并不那么重要。如果这些指标真的那么重要，我们就不会走到今天这个地步。

But I think it's interesting because it's a reference point to compare things today to in the past and not so much to, you know, you must create an ex-valuation or else you're overvalued, but just to compare where we are today compared to where we are back in the dot com boom. A popular way to compare or to value equities is a relative way, just to compare equities to bonds and another popular way is to compare it on a more fundamental way, say price to earnings, price to sales and so forth.

我觉得这很有趣，因为它提供了一个参考点，可以用来比较今天和过去的情况，而不是说你必须创造一个特定的估值，否则就会被高估。这只是为了比较我们今天所处的位置和我们在互联网泡沫时期的位置。一个常见的比较或估值股票的方法是相对估值，比如将股票与债券比较，另一种常见的方法是通过更基本的方式比较，比如市盈率、市销率等等。

So looking at relative, relative valuation, a popular way is to compare, look at something called the equity risk premium, which is say, just thinking at equities as if they were some kind of bond and created some kind of earnings yield based on expected earnings. And then compare that to what you could get if you were to invest in fixed income and this is called an equity risk premium. And if you look at this, it's pretty low and it's comparable to what people compute as the term premium for bonds to be.

从相对估值的角度来看，一个常用的方法是比较所谓的股票风险溢价。简单来说，就是将股票看作某种类型的债券，基于预期收益计算出收益率，然后将其与投资固定收益产品（如债券）所能获得的收益进行比较，这就是股票风险溢价。目前来看，这个溢价相当低，与人们计算出的债券期限溢价类似。

Now all this stuff is model dependent, so it's probably nonsense. Actually, I do believe it's nonsense, but I use it simply because consistently speaking, comparing what it is saying today to comparing to what it was in the past, you can see that now things are quite elevated today compared to where it's been in the past few years.

现在所有这些内容都依赖于模型，因此它可能是无稽之谈。实际上，我确实觉得这是无稽之谈，但我使用它只是因为在一致性上，通过比较它今天所说的内容和过去的一些说法，你可以发现现在的情况比过去几年明显要高。

Another popular way to value equities is to look at price relative to fundamental metric, be it earnings, here let's look at sales because earnings could be subject to all sorts of other accounting adjustments, but sales I think it's kind of a more straightforward way to look at this. On the price to sales ratio, you can see that stocks today are fairly elevated and comparable to errors like the dot com boom. So these are metrics that are now suggesting some sort of fundamental overvaluation.

另一种常见的股票估值方式是观察股价相对于某个基本指标的比例，无论是收益。不过，让我们在这里看看销售额，因为收益可能会受到各种会计调整的影响，而销售额则是一个更直接的观察方式。通过价格与销售额的比率，你可以看到当前的股票价格相对较高，与互联网泡沫时期相当。因此，这些指标现在暗示着某种程度上的基础估值过高。

Now you can also look at anecdotal reports. We do have some obvious bubbles in the markets. Let's say some stocks, storied AI related stocks that trade a P to say 600 and just kind of seem to go higher and higher and higher. Now let's be totally clear. This is a company that has significant growth in earnings, but let's also be totally clear that there's no way you can lower your going earnings such as the P ratio of 600 makes sense. It's obvious, obvious, total complete bubble.

现在，你也可以看看一些轶事报道。显然，市场中存在一些泡沫。比如，有些与人工智能相关的热门股票，它们的市盈率高达600倍，并且似乎在不断上涨。我们需要明确指出，这家公司确实有显著的盈利增长，但同样需要明确的是，这样高的市盈率绝对不合理。这显然是一个彻头彻尾的泡沫。

Again, you can also have a lot of anecdotal reports, tremendous amounts of retail enthusiasm and retail traders have been very successful. Buying the dip has worked very, very well and they've all done very well. But just speaking of someone who has played this game for a long time, one of the weaknesses that you learn overall in time is that when you continue to win, you have an emotional impulse to bit bigger and bigger and become a year and a year and that continues to work until it doesn't.

再说一次，你会发现有很多轶事报告，大量的零售投资者充满热情，零售交易者非常成功。“逢低买入”策略一直以来效果很好，他们都取得了不错的成绩。但作为一个玩这个游戏很长时间的人，我想说你会慢慢认识到一个弱点：当你持续获胜时，你会有一种情绪上的冲动去下注越来越大，这种策略会一直有效，直到某天它不再奏效为止。

It's something that we kind of learned throughout our trading journey that everyone walks on. And so when you have more and more retail involvement, that often tells me that people would become more and more levered and become more and more vulnerable to snapbacks. And of course, some of the signs are you have fabled 23-year-old hedge fund managers making a killing, managing billions.

在我们的交易旅程中，我们慢慢学到了一些道理，就是每个人都会经历的。当越来越多的散户参与进来时，这通常告诉我，人们会变得越来越依赖杠杆，从而在市场快速反弹时变得更加脆弱。当然，有些迹象表明了这一点，比如传闻中的23岁对冲基金经理赚得盆满钵满，管理着数十亿美元的资金。

And to be clear, I'm sure this person is very high IQ, but at the end of the day, you're 23 years old. So I think these are all signs of, there's a lot of exuberance in the markets and does the mean can go higher, but it does mean personally speaking that this will inevitably be followed by some kind of bust. Don't know when, but it will happen.

需要说明的是，我确信这人智商非常高，但说到底，你才23岁。因此，我认为这显示了市场上有很多过度乐观的迹象。虽然这可能意味着市场会继续上涨，但就我个人而言，我认为这不可避免地会被某种程度的崩溃所跟随。虽然我不知道何时会发生，但这一定会发生。

All right, so that's all I prepared for today. Next week, not much going on, headline will be on Jackson Hole, where Paul will speak and maybe say something about monetary policy, but maybe with so much uncertainty, so much data to be coming. And of course, disagreement on the FOMC, my best is probably not too interesting. All right, talk to you all then.

好的，这就是我今天准备的全部内容。下周没有太多事情，重点将是杰克逊霍尔会议，保罗将在那里发言，可能会谈到一些货币政策。但由于存在很多不确定性，还有大量数据即将发布，当然，美联储公开市场委员会（FOMC）内部也有分歧，我认为可能不会太有趣。好了，到时候再聊。