Don’t Blame Robots for Low Wages
By Paul Krugman

The other day I found myself, as I often do, at a conference discussing lagging wages and soaring inequality. There was a lot of interesting discussion. But one thing that struck me was how many of the participants just assumed that robots are a big part of the problem — that machines are taking away the good jobs, or even jobs in general. For the most part this wasn’t even presented as a hypothesis, just as part of what everyone knows.

And this assumption has real implications for policy discussion. For example, a lot of the agitation for a universal basic income comes from the belief that jobs will become ever scarcer as the robot apocalypse overtakes the economy.

So it seems like a good idea to point out that in this case what everyone knows isn’t true. Predictions are hard, especially about the future, and maybe the robots really will come for all our jobs one of these days. But automation just isn’t a big part of the story of what happened to American workers over the past 40 years.

We do have a big problem — but it has very little to do with technology, and a lot to do with politics and power.

Let’s back up for a minute, and ask: What is a robot, anyway? Clearly, it doesn’t have to be something that looks like C-3PO, or rolls around saying “Exterminate! Exterminate!” From an economic point of view, a robot is anything that uses technology to do work formerly done by human beings.

And robots in that sense have been transforming our economy literally for centuries. David Ricardo, one of the founding fathers of economics, wrote about the disruptive effects of machinery in 1821!

[Paul Krugman did explanatory journalism before it was cool, moving from a career as a world-class economist to writing hard-hitting opinion columns. For an even deeper look at what’s on his mind, sign up for his weekly newsletter.]

These days, when people talk about the robot apocalypse, they don’t usually think of things like strip mining and mountaintop removal. Yet these technologies utterly transformed coal mining: Coal production almost doubled between 1950 and 2000 (it only began falling a few years ago), yet the number of coal miners fell from 470,000 to fewer than 80,000.

Or consider freight containerization. Longshoremen used to be a big part of the scene in major port cities. But while global trade has soared since the 1970s, the share of U.S. workers engaged in “marine cargo handling” has fallen by two-thirds.

Technological disruption, then, isn’t a new phenomenon. Still, is it accelerating? Not according to the data. If robots really were replacing workers en masse, we’d expect to see the amount of stuff produced by each remaining worker — labor productivity — soaring. In fact, productivity grew a lot faster from the mid-1990s to the mid-2000s than it has since.

So technological change is an old story. What’s new is the failure of workers to share in the fruits of that technological change.

I’m not saying that coping with change was ever easy. The decline of coal employment had devastating effects on many families, and much of what used to be coal country has never recovered. The loss of manual jobs in port cities surely contributed to the urban social crisis of the ’70s and ’80s.

But while there have always been some victims of technological progress, until the 1970s rising productivity translated into rising wages for a great majority of workers. Then the connection was broken. And it wasn’t the robots that did it.

What did? There is a growing though incomplete consensus among economists that a key factor in wage stagnation has been workers’ declining bargaining power — a decline whose roots are ultimately political.

Most obviously, the federal minimum wage, adjusted for inflation, has fallen by a third over the past half century, even as worker productivity has risen 150 percent. That divergence was politics, pure and simple.

The decline of unions, which covered a quarter of private-sector workers in 1973 but only 6 percent now, may not be as obviously political. But other countries haven’t seen the same kind of decline. Canada is as unionized now as the U.S. was in 1973; in the Nordic nations unions cover two-thirds of the work force. What made America exceptional was a political environment deeply hostile to labor organizing and friendly toward union-busting employers.

And the decline of unions has made a huge difference. Consider the case of trucking, which used to be a good job but now pays a third less than it did in the 1970s, with terrible working conditions. What made the difference? De-unionization was a big part of the story.

And these easily quantifiable factors are just indicators of a sustained, across-the-board anti-worker bias in our politics.

Which brings me back to the question of why we’re talking so much about robots. The answer, I’d argue, is that it’s a diversionary tactic — a way to avoid facing up to the way our system is rigged against workers, similar to the way talk of a “skills gap” was a way to divert attention from bad policies that kept unemployment high.

And progressives, above all, shouldn’t fall for this facile fatalism. American workers can and should be getting a much better deal than they are. And to the extent that they aren’t, the fault lies not in our robots, but in our political leaders.

諾貝爾獎得主克魯曼：工資低落別怪機器人！

近年來每當觸及工資低迷議題，評論矛頭往往指向機器人搶走工作。這種論調儼然已蔚為常識，進而影響政策措施，比方說提倡「全民基本收入」的運動，就是因為認定工作只會愈來愈稀少、機器人接掌經濟的末日災難將至。

但2008年諾貝爾經濟學獎得主克魯曼駁斥這種論調並非事實。他在15日紐約時報專欄撰文指出，回顧過去40年來，美國勞工之所以遭受工作流失和工資低落，最主要因素不是自動化，而是與政治和權力更有關係。

先界定何謂「機器人」（robot）。機器人外觀未必像電影「星際大戰」裡的人形機器人C-3PO，克魯曼說，從經濟觀點來看，凡是「能運用科技完成先前人工工作」之物，都可稱為機器人。

就這個定義而論，機器人改造經濟已長達數世紀之久。早在1821年，古典經濟學家李嘉圖（David Ricardo）就已撰文探討機械帶來的破壞性效應。

露天採礦和山頂移平等技術促成煤礦開採徹底轉型，導致1950年至2000年煤礦生產幾乎倍增，但採煤機械化也導致煤礦工從47萬人降到如今不到8萬人。碼頭裝卸工曾活躍各大港市，但隨著貨運貨櫃化，1970年代迄今全球貿易激增的同時，從事「船貨處理」的美國勞工數已銳減三分之二。

由此看來，科技帶來的破壞性效應並不是新的現象。話雖如此，自動化取代人力的趨勢是否正在加速？克魯曼表示，從數據研判，並沒有。若是機器人正大量取代勞工，剩餘勞工每人的生產的東西數量（即勞工生產力）理應飆升才對；事實上，1990年代中期至2000年代中期的生產力增幅，遠大於2000年代中期至今的增幅。

克魯曼指出，科技帶來改變是舊聞，新的發展是：勞工未能分享到科技進步的果實。

科技進步的過程總是有受害者，煤礦業就業人數下降已重創許多家庭，許多工作一去而不復返，港市勞力工作流失助長1970年代和1980年代美國都市社會危機。但之前生產力提升通常會轉化為大多數勞工的工資上漲，直到1970年代後這種關聯性才被打破。這可不是機器人的錯。

那是誰的錯？克魯曼說，愈來愈多經濟學家認為，造成工資停滯的一個關鍵因素是勞工議價力式微，而這是肇因於政治因素。

他指出，過去50年來，勞工生產力激增150%，但經通膨調整的美國聯邦最低工資卻下降三分之一。為何分歧這麼大？顯然是政治因素使然。

克魯曼解釋，工會式微，從1973年美國民間企業勞工有25%是工會勞工，到如今降到只有6%，反映美國政治環境對勞工籌組工會敵意甚深，對雇主卻很友善。

工會式微已造成巨大影響。以卡車貨運業為例，卡車司機昔日是待遇好的工作，但現在酬勞比1970年代縮水三分之一，而且工作條件惡劣。去工會化（de-unionization）是主因。

克魯曼認為，現在所謂「機器人末日」（robot apocalypse）之說，是一種轉移注意力的伎倆—避免正視美國現行制度被操弄成不利勞工的問題。這跟藉由大談「技能缺口」（skills gap），企圖轉移各界對爛政策導致失業人數居高不下的注意，如出一轍。

他呼籲進步改革人士不應陷入這種看似理所當然的宿命論思維。勞工有能力、也應該獲得更好的待遇；若是沒有，應該責怪政治領導人，不要錯怪機器人。

原文參照：
https://www.nytimes.com/2019/03/14/opinion/robots-jobs.html

2019-03-15 經濟日報 記者湯淑君/即時報導