Though 2017 was a great year for Silicon Valley Bank loan and deposit growth, we believe that Silicon Valley Bank is entering a period of lower returns from increasing competition and declining credit quality. Recently, management has said the company has benefited from sovereign wealth funds and SoftBank's aggressive investment campaign. We suspect these atypical investors represent the last of the few remaining pillars of funding supporting venture capital activity.
One concerning trend of this tech cycle is the use of leverage by venture-capital-backed startups. Unlike in 2001, where equity was often the sole funding source, venture debt has played an increasing role at early-stage companies. Rather than dilute its equity stake, Uber has raised more than $3 billion in debt from major banks. WeWork, another unicorn with an unproven business model, has reportedly sought to obtain a $750 million credit line. Industrywide, we believe that as a result of rising competition, there has been a decline in underwriting quality by providing startups debt earlier in their life cycle. Too frequently, we have observed Silicon Valley Bank's public borrowers do not generate cash or have the asset coverage to take on significant debt. Yet, Silicon Valley Bank has been willing to provide them with loans.
Given that Silicon Valley Bank grew loans at a 20% compound annual rate the past five years in a venture debt market that has become highly competitive, we believe a significant rise in charge-offs accompanied by slower growth is inevitable. We forecast charge-offs will increase to almost 2% in 2019 and 2020 and decline thereafter, though we'd warn this could be substantially higher. We don't think the bank's loans are as good as in previous cycles, given industrywide trends. We also believe the bank will see significant negative loan growth in a venture capital downturn. In past downturns, loans have contracted by as much as 18%. Silicon Valley Bank has experienced tremendous growth in capital call lines, which could lead to loan balances falling farther than they have in previous downturns. We expect to see loans contract by 20%-25% from peak levels over our forecast period.
Despite having an impressively cheap source of commercial deposits, in our view, SVB Financial does not possess a moat. We do not expect Silicon Valley Bank to generate significant excess returns. Historically, banks have established moats through cost advantages, customer loyalty through high switching costs, or structural advantages in the banking system in which they operate. In 2017, 80% of Silicon Valley Bank's deposits were non-interest-bearing, and we calculated a total cost of deposits at less than 2 basis points annually, an extremely low number. In comparison, Wells Fargo, a bank that's more than 10 times the size and achieved a wide moat partly through cheap retail deposits, has a total cost of deposits of less than 8 basis points. In addition, First Republic, one of SVB's closest competitors, has a cost of deposits exceeding 14 basis points. While SVB does enjoy a cheap source of deposits, we believe there's a substantial opportunity cost attached to its inexpensive commercial deposits. Unlike its competitors and most banks, Silicon Valley Bank's depositors are not retail deposits, but are rather concentrated among startups and the private equity and venture capital funds that invest in them. As a result, SVB Financial is a bank that is highly leveraged to the Silicon Valley funding cycle, and its deposit base can fluctuate wildly. SVB's funding cost may be very low on an absolute basis, but movement in these commercial deposits is difficult to forecast, given startups' tendency to burn cash while routinely needing to raise new capital. These are unlike retail deposits, which tend to be very stable and moaty. From the end of 2008 to 2010, as startups became the hot investment, deposits grew approximately 50% annually. In 2001, after the tech recession, deposits dropped more than 30%. Given this variability in deposits, the bank must keep a substantial portion of its deposits in cash and securities. In 2017, the bank held more than 61% of its on-balance-sheet deposits in cash and securities. In comparison, First Republic keeps only 24% of its deposits in cash and securities. In addition, a flood of new deposits can result in more cash than a bank can effectively deploy, which we believe has happened. From an operational perspective, the bank enjoys an efficiency ratio of 51.1%, which is respectable.
SVB Financial maintains that its long history and strong relationships within the Silicon Valley community provide a competitive edge in lending to private equity and venture capital firms and entrepreneurs. We believe there is some validity to this statement. Silicon Valley Bank lends significantly to private equity and venture capital funds in the form of capital call lines of credit. These are short-term loans that enable these funds to delay calling capital from investors, which increases the internal rate of return these funds report to their investors. For example, if a private equity firm raises capital for a fund lasting five years and uses a capital call line of credit provided by Silicon Valley Bank for the first six months of the investment, it enables it to reduce the time and the cost of capital used to calculate its internal rate of return. Usually, these lines of credit are short-term loans ranging from three to six months and are backed by investor commitments. These loans are repaid as soon as the private equity or venture capital fund calls capital from its investors. These loans have grown rapidly at Silicon Valley Bank. At the end of 2017, private equity and venture capital loans accounted for more than 40% of the total loan book, up from 15% in 2012. The company trumpets the low-risk nature of these loans, given their short-term nature. While these loans have been profitable, they are also highly transactional and depend on deal activity. If deal activity continues to slow, it is almost certain that the bank would see significant negative loan growth, something few investors are anticipating.
Throughout much of the bank's history, SVB Financial has been able to generate returns in excess of its cost of capital by successfully lending to venture-capital-backed companies, often profitless firms that are shunned by many traditional lenders. These are companies that likely have little in the way of assets, are generating losses, possibly have an undefined business model, and whose ability to repay depends on reaching an initial public offering or being acquired. As a result, SVB was able to charge borrowers high rates of interest and attach equity warrants to their loans, giving Silicon Valley Bank significant upside should a borrower be acquired or complete an IPO. In 2017, gains on warrants accounted for almost 6% of the bank's pretax income. SVB maintains that its unique position within the Silicon Valley ecosystem has provided it with better information to identify which companies will generate significant growth and possibly even achieve an IPO or be acquired. Before 2009, Silicon Valley Bank enjoyed little competition and had the ability to pick and choose to whom they lent by identifying the best early-stage companies sooner than its peers. Until recently, this segment of lending was a small niche dominated by a few small banks. However, we believe this has changed as new lenders have entered the venture debt market, putting pressure on the rates and terms Silicon Valley Bank can offer. Today, Silicon Valley Bank is competing against large lenders like Comerica, business development companies, hedge funds, and even nonfinancial startups to provide funding to companies. Simply put, our view is that Silicon Valley Bank is taking equitylike risk, but earning bondlike returns. Going forward, we believe increased competition will be reflected in charge-offs, negative growth in loans and deposits, and a higher efficiency ratio. We believe this increased competition supports our thesis that SVB Financial lacks a moat.
SVB Financial operates in the U.S. banking system, which we assess as fair from a stability perspective. Though regulation has become considerably stronger in the past several years, the country still uses a complex and somewhat archaic system of regulation. Furthermore, the company's banking market is quite fragmented; Silicon Valley Bank must compete with a variety of regional and community banks, as well as large money-center institutions. Over the past 50 years, the banking system has achieved returns in line with its cost of capital, which supports our view of the environment as intensely competitive. Our outlook is more positive from a macroeconomic and political standpoint. The U.S. is still the world's leading democracy, has increased GDP at a steady pace for years, and maintains the world's reserve currency, all of which contribute to banking stability.
We are raising our fair value estimate to $98 per share from $93, due to an increase in the time value of money. As of June 30, 2018, our fair value estimate is about 1.1 times book value per share. We still believe that Silicon Valley Bank's commercial loan portfolio is of low quality and are convinced credit performance will eventually deteriorate. We now forecast the bank to more gradually shrink its loan and deposit portfolio, rather than a sudden downturn. That said, when the venture capital market turns, we don't expect the changes in the bank's balance sheet or deposit base to be gradual or modest. We still think in five years, SVB Financial will be a smaller bank in a more competitive market.
Over the past decade, venture capital deal activity has grown at a compounded rate of 9%. More recently, deal activity has grown by more than 20% on average over the past five years. If we assume sustainable growth in venture capital activity was closer to 9% starting in 2014 (with excess growth representing a bubble in activity), and that SVB Financial's portfolio growth mirrors total venture capital deal value, that would imply that the bank's loan portfolio would need to shrink by 25% over the next five years. Over the next five years, we therefore assume loans fall by about 24%; however, this could happen more quickly than we expect. We anticipate deposits would shrink by about 20% as well. It is likely the bank will search for higher-cost deposits elsewhere in the event of a declining core customer deposit base.
We're forecasting that net charge-offs peak in 2020 at about 2.2% of total loans. In previous peaks, net charge-offs reached 3.3% and 2.5% in 2009. Today, the bank is more tilted toward private equity capital call lines, which should perform better in the event of a downturn. That said, we think the rest of SVB's loan portfolio is materially worse than in previous cycles.
We still believe SVB will benefit from rising interest rates, but we doubt that the bank will enjoy this cheap of funding if deposit growth turns negative. The bank will likely have to go out and pay up for deposits. In light of this, we expect net interest margins to increase only 170 basis points to 4.9% in 2022 from 3.2% in 2017.
The greatest risk to SVB would be a prolonged slowdown in exit markets and venture capital and private equity funding. This would likely result in significantly increased charge-offs and allowances for loan losses. We do worry that this most recent tech cycle, unlike previous cycles, has been partly fueled by leverage. Charge-offs could very well exceed previous slowdowns in funding. In 2000, charge-offs peaked at 3.33% of average loans. In addition, the bank's deposits can be highly volatile at the beginning and during the emergence of recessions. In the first quarter of 2001, deposits declined more than 30% year over year and rebounded 38% at the end of 2008. While the bank enjoys a low-cost deposit base, 50% of which belong to early-stage companies, we worry how sustainable many of those businesses are, given their unproven business models and tendency to burn cash. Should the bank experience a significant reduction in deposits, it may need to pay up to acquire new types of deposit funding. This would likely result in a significant reduction in net interest margin. We also believe a slowdown in the innovation economy would cause a significant reduction in fee income, as the bank's noninterest income is derived from startup economy.
We view SVB's stewardship of shareholder capital as Poor. Given the flood of investment into Silicon Valley and increased competition to provide companies with venture debt, we would prefer a bank to operate with more skepticism and be willing to walk away from deals. Greg Becker has been CEO of the company since April 2011. Under his leadership, loans have grown from $5.6 billion to $23 billion, a compounded growth rate of more than 23%, leading us to believe that he and his team have walked away from few, if any, deals. In comparison, Becker's predecessor Ken Wilcox, of whom we have a favorable opinion, increased loans at an annual rate of only 12%. Unlike the late 1990s, when companies were coming to market with little to no debt, this tech cycle seems to be more enamored with leverage, and SVB has been an enthusiastic participant. Jet.com, a highly profitable loan made by this management team, raised $565 million in equity capital. In addition to that equity, Jet.com received at least $125 million in debt finance, at least some of which came from Silicon Valley Bank. While the investment worked out well, we believe it entailed a substantial amount of risk given the amount of debt, the questionable business model, and the rate at which we suspect the startup was burning cash. It is unlikely that management could have known in September of 2014 when the loan was made that it would result in an acquisition by Walmart. In comparison, at the end of 1997, Amazon had about $76 million in debt, but it was only after the company went public do we find any debt in its public filings. EBay, another tech darling of that era, didn't have any debt at the time of its IPO. In contrast, today's darling, Uber, has raised over $3 billion in debt, though we are not aware of any involvement by SVB. In addition, another unicorn, WeWork, is reported to have sought a $750 million credit line in November 2015. However, unlike Uber, WeWork has an unproven business model, providing short-term office leases to fledgling companies. We are not aware of SVB providing any loans to WeWork. However, we believe this demonstrates that this tech cycle has partly been fueled by debt. It is our opinion that management has not been shy about providing credit to increasingly shaky credits in the innovation economy and supports our opinion that the company has been a poor steward of capital.
In addition, the bank has boasted about its global reach and ability to secure a Chinese banking license to provide banking services denominated in Chinese yuan in mainland China. At best, we believe that Chinese expansion lies far outside this bank's circle of competence and is a waste of management's time and shareholders' capital. At worst, we believe this is promotional and exemplifies the bank's willingness to take risk without generating a commensurate return.
