What to do with cash: reinvest, buy back or hold

1 The bottom line

Capital allocation is the CEO's core job, and the best use of cash is whatever earns the highest return now. Cash left to habit or short-term earnings pressure quietly destroys value.

2 Key takeaways

• Own capital allocation yourself and give it most of your attention. Most leaders arrive with no training in it.
  • Henry Singleton, Teledyne1:45, spent his time deploying cash and returned 20.4% a year for almost 30 years.
• Reinvest first where the return can be measured. A known payback beats a hopeful one.
  • Howard Schultz, Starbucks43:50, held every new store to $1 million of year-one sales on a $500,000 investment.
  • Morris Chang, TSMC1:47:33, put $17 billion of $20 billion operating profit into capex in 2020.
• Buy back shares only when they are clearly cheap. The same cash earns more in your own stock at the right price than almost anywhere else.
  • Singleton, Teledyne10:27, retired about 90% of shares over eight tenders, and holders earned a 42% compound annual return.
  • Frank Blake, Home Depot3:41:20, retired 30% of shares, mostly at $30 to $50, against $340 today.
• Protect long-horizon spending from the pressure to look profitable. Cutting investment lifts profit now and bills you later.
  • Morris Chang, TSMC41:30, fixed R&D at 8% of revenue regardless of recession.
• Size big bets by the payoff, and sell what does not fit the core. Cash tied up in a distraction cannot fund the real engine.
  • Blake, Home Depot3:41:20, sold HD Supply for about $8.3 billion and bought back 14% of shares in the first year.
  • Mark Zuckerberg, Meta5:15:54, has lost about $60 billion on Reality Labs since 2019.

I think we can earn a better return buying our shares at these levels than by doing almost anything else.

Henry Singleton, Teledyne

2.1 Make capital allocation the CEO's job

The person at the top decides where cash goes, so that person must learn the skill and spend time on it.

• Spend your own time deploying the cash your businesses generate, as Henry Singleton, Teledyne1:45, did.
  • After 1967 he stepped back from operations once George Roberts became president.
  • He judged success by per share value, not size or growth.
  • Teledyne returned 20.4% a year for almost 30 years, and Buffett said Singleton had the best operating and capital deployment record in American business.
• Admit that most CEOs lack this skill and build it on purpose Singleton, Teledyne1:45.
  • Most CEOs rise through marketing, engineering or politics and arrive with no capital allocation experience.
  • His line: "The heads of many companies are not skilled in capital allocation."
• When a big payout lands, decide at once where it goes. Elon Musk, Zip25:55, did.
  • Compaq paid $37 million in cash for Zip2, leaving him $22 million at 27.
  • He put almost all of it into the next company, saying he would be either wealthy or broke.
  • He repeated the move later by putting Tesla stock cash into Twitter.

2.2 Reinvest where the return is measurable

Reinvestment wins when you can state the payback before you spend.

• Set a store-level return model and hold every new site to it, as Howard Schultz, Starbucks43:50, did.
  • He chose the first 500 locations himself.
  • Each had to hit year-one sales of $1 million on a $500,000 investment and an operating profit above 20%.
  • New stores paid back in about two years, sometimes a year and a half. The model rested on a roughly 80% gross margin and loyal repeat visits.
• Rank uses of space or money by return on the scarce resource, as Joe Coulombe, Trader Joe's2:54:56, did.
  • Shelf inches are scarce, so he ranked items by dollars of margin per shelf space, not margin percent.
  • A $20 item earning a couple of dollars beats a $4 item earning a dollar if it takes the same room.
  • Sales per square foot topped $2,000, versus about $1,200 at Costco and about $1,000 at Whole Foods.
• When adding capacity stops paying, squeeze what you already own. Frank Blake, Home Depot3:34:10, did.
  • In 2008 he froze expansion, wrote off about $1 billion in pipeline and closed around 30 stores.
  • Sales per store rose from about $30 million to about $65 million over 11 years.
  • Revenue grew from $70 billion to $130 billion and net income from $4 billion to $11 billion with a flat store count.
• Where the field is a treadmill, plow nearly all operating profit back, as Morris Chang, TSMC1:47:33, did.
  • In 2020 TSMC earned $20 billion of operating profit on $48 billion of revenue and spent $17 billion on capex.
  • The goal was to keep everybody else five plus years behind.
  • It raised the plan to $30 billion for 2021 and $100 billion over three years.
  • Around 2010 it committed $9 billion and 6,000 employees to a dedicated plant for Apple1:39:55, built in 11 months.

2.3 Protect long-term spending from short-term earnings

Skipping investment makes this year's profit look good by spending assets built in the past.

• Refuse to cut investment to look profitable, as Akio Morita, Sony1:00:00, did.
  • He treated marketing a new product as a return that arrives in five or 10 years.
  • He told a hesitant president to spend a million or two million dollars within two months or be fired.
  • He said managers can look good on the bottom line while destroying the company by failing to invest.
• Resist Wall Street's quarterly focus on earnings, like David Packard, Hewlett-Packard35:35.
  • He noted that profits can always be lifted for a while by cutting new product design and engineering.
  • His view: "good new products are the lifeblood of technical companies."
• Fix the R&D budget as a share of revenue so it stops being a yearly fight, as Morris Chang, TSMC41:30, did.
  • At Texas Instruments he was refused every time he asked to move R&D from 4.8% to 5.5%.
  • Back at TSMC in 2009, R&D ran at 6 or 7% and was renegotiated yearly, so he picked 8% regardless of recession.
  • The R&D director called it the best thing done for his team, and it fed the push to 28nm.
• Treat advertising as an investment, not an expense, as Ray Kroc, McDonald's38:45, did.
  • The McDonald brothers refused to give 1% of sales to a shared campaign.
  • Kroc called himself the promoter, who sees money coming back with interest.
  • He approved a $180,000 California TV campaign but refused to fund it with a one cent price rise.
• Split resources between the present and the future, as Ingvar Kamprad, IKEA54:10, did.
  • Half went to improving what exists and half to what comes next.
  • IKEA is growing faster today than during a roughly 20 year flat stretch.

2.4 Buy back shares when they are cheap

Repurchases are a real use of cash only when price is well below value, and they pair well with selling what does not belong.

• Buy your own stock when it is clearly too cheap, even if the market reads it as weakness. Henry Singleton, Teledyne10:27, did.
  • In 1972 he judged the stock too cheap and announced a tender offer.
  • Over eight tenders to 1984 he retired about 90% of the shares.
  • Shareholders earned a 42% compound annual return across the tenders.
• Sell the side business and send the cash to buybacks. Frank Blake, Home Depot3:41:20, did.
  • HD Supply had reached 13% of revenue by 2006, and Blake judged it a different model that distracted from the core.
  • He sold it in 2007 for about $8.3 billion and bought back 14% of shares in the first year.
  • From 2008 to 2012 the stock rose 132%, then 33% in 2013, 27% in 2014 and 26% in 2015.
• Keep buying the stock of the business you know best and never sell it. John D. Rockefeller, Standard Oil36:53, did.
  • He bought even during the crude boycott, when employment fell from 1,200 to 70 and partners panicked.
  • His rule: "Let it feed upon itself."
  • His net worth rose by $100 million between about 1905 and 1915 by doing nothing.

2.5 Buy stakes and inputs at the right price, and avoid vanity spend

Outside the core, buy only what is cheap and understood, and keep spending tied to the product.

• Prefer stakes in public companies to overpriced whole companies. Singleton, Teledyne34:51, did.
  • A stock at a multiple of six or seven costs 12 or 14 times earnings once you pay for control.
  • Teledyne's insurance subsidiaries held over 70% of their equities in five companies, with 25% in Litton.
  • By 1977 Teledyne was the largest shareholder in nine Fortune 500 companies, and Singleton37:47 never sought a board seat or control.
• Buy your key input in large lots when its price bottoms. Rockefeller, Standard Oil21:20, did.
  • He sent John Andrews to Oil City with that single job.
  • He borrowed mostly to buy crude when prices fell, though it made everyone nervous.
  • His view: "We will surely make a great mistake if we do not buy."
• Take long-shot bets only when the upside is huge, and count the cost. Jeff Bezos, Amazon2:31:10, argued for this.
  • He said a 10% chance of a hundred times payoff is a bet to make every time.
  • AWS grew to a roughly $80 billion run rate from a four to six billion dollar business in 2014.
  • The counterweight is Meta5:15:54, where Reality Labs lost about $60 billion since 2019 and napkin math keeps cumulative cash flow negative until at least 2035.
• Spend on what touches the product, not on show. Akio Morita, Sony1:02:55, held this line.
  • His rule: "You need to air condition your factories before your offices."
  • By contrast Steve Jobs, NeXT9:15, paid $100,000 for a logo and bought $2,000 chairs for a company with no product, while his burn rate climbed year after year.

2.6 Where they differ

The right use of cash depends on your funding, your industry and who owns you.

• Reinvesting everything works until lenders balk.
  • Phil Knight, Nike29:15, kept no idle cash and doubled his order each time he repaid the bank, so sales doubled five straight years.
  • Banks repeatedly cut him off for lack of cash balances, and the company came near collapse.
• Payouts are a real dispute even inside one company.
  • Rockefeller, Standard Oil54:00, preferred reinvesting, but his executives raised the dividend from about 11% to 31% in 1897 and 33% in 1899.
  • Standard paid out $250 million from 1893 to 1901, and the share price rose from $176 in 1896 to $458 three years later.
• Capital-heavy models carry a higher cost of capital than float-funded ones.
  • In retail, Amazon2:25:50 collects from customers at once and pays suppliers later. In cloud it must buy servers and build data centers up front.
  • Capital leases make data center costs pay as you go, but free cash flow still turned sharply negative, versus $27 billion in the twelve months to June 2020.
• Patient, half-and-half spending suits owners who are not judged quarterly.
  • IKEA54:10 had a roughly 20 year stretch of flat growth while profitable, which a public company could not have afforded.