The numbers for our key metrics, which include our daily active users (DAUs), monthly active users (MAUs), and average revenue per user (ARPU), are calculated using internal company data based on the activity of user accounts. While these numbers are based on what we believe to be reasonable estimates of our user base for the applicable period of measurement, there are inherent challenges in measuring usage of our products across large online and mobile populations around the world. In addition, we are continually seeking to improve our estimates of our user base, and such estimates may change due to improvements or changes in our methodology. In 2016, we estimate that "duplicate" accounts (an account that a user maintains in addition to his or her principal account) may have represented approximately 6% of our worldwide MAUs. We also seek to identify "false" accounts, which we divide into two categories: (1) user-misclassified accounts, where users have created personal profiles for a business, organization, or non-human entity such as a pet (such entities are permitted on Facebook using a Page rather than a personal profile under our terms of service); and (2) undesirable accounts, which represent user profiles that we determine are intended to be used for purposes that violate our terms of service, such as spamming. In 2016, for example, we estimate user-misclassified and undesirable accounts may have represented approximately 1% of our worldwide MAUs. We believe the percentage of accounts that are duplicate or false is meaningfully lower in developed markets such as the United States or United Kingdom and higher in developing markets such as India and Turkey. However, these estimates are based on an internal review of a limited sample of accounts and we apply significant judgment in making this determination, such as identifying names that appear to be fake or other behavior that appears inauthentic to the reviewers. Our estimates may change as our methodologies evolve, including through the application of new technologies, which may allow us to identify previously undetected false or duplicate accounts and improve our ability to evaluate a broader population of our users. As such, our estimation of duplicate or false accounts may not accurately represent the actual number of such accounts. 

Our data limitations may affect our understanding of certain details of our business. For example, while user-provided data indicates a decline in usage among younger users, this age data is unreliable because a disproportionate number of our younger users register with an inaccurate age. Accordingly, our understanding of usage by age group may not be complete.

In addition, our data regarding the geographic location of our users is estimated based on a number of factors, such as the user's IP address and self-disclosed location. These factors may not always accurately reflect the user's actual location. For example, a user may appear to be accessing Facebook from the location of the proxy server that the user connects to rather than from the user's actual location. The methodologies used to measure user metrics may also be susceptible to algorithm or other technical errors. Our estimates for revenue by user location and revenue by user device are also affected by these factors. For example, in late 2015, we discovered an error in the algorithm we used to attribute our revenue by user geography in late 2015.geography. While this issue did not affect our overall worldwide revenue, it did affect our attribution of revenue to different geographic regions. The fourth quarter of 2015 revenue by user geography and ARPU amounts were adjusted to reflect this reclassification.

We regularly review our processes for calculating these metrics, and from time to time we may discover inaccuracies in our metrics or make adjustments to improve their accuracy, including adjustments that may result in the recalculation of our historical metrics. We believe that any such inaccuracies or adjustments are immaterial unless otherwise stated. In addition, our DAU and MAU estimates will differ from estimates published by third parties due to differences in methodology.

The numbers of DAUs and MAUs discussed in this Quarterly Report on Form 10-Q, as well as ARPU, do not include Instagram, WhatsApp, or Oculus users unless they would otherwise qualify as such users, respectively, based on their other activities on Facebook. In addition, other user engagement metrics included herein do not include Instagram, WhatsApp, or Oculus unless otherwise specifically stated.

The accompanying unaudited condensed consolidated financial statements have been prepared in accordance with generally accepted accounting principles in the United States (GAAP) and applicable rules and regulations of the Securities and Exchange Commission regarding interim financial reporting. Certain information and note disclosures normally included in the financial statements prepared in accordance with GAAP have been condensed or omitted pursuant to such rules and regulations. As such, the information included in this quarterly report on Form 10-Q should be read in conjunction with the consolidated financial statements and accompanying notes included in our Annual Report on Form 10-K for the fiscal year ended December 31, 2016.

The condensed consolidated balance sheet as of December 31, 2016 included herein was derived from the audited financial statements as of that date, but does not include all disclosures including notes required by GAAP.

The condensed consolidated financial statements include the accounts of Facebook, Inc. and its wholly owned subsidiaries. All intercompany balances and transactions have been eliminated.

The accompanying condensed consolidated financial statements reflect all normal recurring adjustments necessary to present fairly the financial position, results of operations, and cash flows for the interim periods, but are not necessarily indicative of the results of operations to be anticipated for the full year ending December 31, 2017.

There have been no changes to our significant accounting policies described in our Annual Report on Form 10-K for the fiscal year ended December 31, 2016 that have had a material impact on our condensed consolidated financial statements and related notes.

Conformity with GAAP requires the use of estimates and judgments that affect the reported amounts in the condensed consolidated financial statements and accompanying notes. These estimates form the basis for judgments we make about the carrying values of our assets and liabilities, which are not readily apparent from other sources. We base our estimates and judgments on historical information and on various other assumptions that we believe are reasonable under the circumstances. GAAP requires us to make estimates and judgments in several areas, including, but not limited to, those related to revenue recognition, collectability of accounts receivable, contingent liabilities,loss contingencies, fair value of financial instruments, fair value of acquired intangible assets and goodwill, useful lives of intangible assets and property and equipment, and income taxes. These estimates are based on management's knowledge about current events and expectations about actions we may undertake in the future. Actual results could differ materially from those estimates.

We compute earnings per share (EPS) of Class A and Class B common stock using the two-class method required for participating securities. We consider restricted stock awards to be participating securities because holders of such shares have non-forfeitable dividend rights in the event of our declaration of a dividend for common shares.

Undistributed earnings allocated to participating securities are subtracted from net income in determining net income attributable to common stockholders. Basic EPS is computed by dividing net income attributable to common stockholders by the weighted-average number of shares of our Class A and Class B common stock outstanding, adjusted for outstanding shares that are subject to repurchase.

For the calculation of diluted EPS, net income attributable to common stockholders for basic EPS is adjusted by the effect of dilutive securities, such as awards under our equity compensation plans and inducement awards under separate non-plan restricted stock unit (RSU) award agreements. In addition, the computation of the diluted EPS of Class A common stock assumes the conversion of our Class B common stock to Class A common stock, while the diluted EPS of Class B common stock does not assume the conversion of those shares to Class A common stock. Diluted EPS attributable to common stockholders is computed by dividing the resulting net income attributable to common stockholders by the weighted-average number of fully diluted common shares outstanding.

Basic and diluted EPS are the same for each class of common stock because they are entitled to the same liquidation and dividend rights.

The numerators and denominators of the basic and diluted EPS computations for our common stock are calculated as follows (in millions, except per share amounts): 

The following table sets forth the cash and cash equivalents, and marketable securities (in millions):

The following table summarizes, for assets or liabilities measured at fair value, the respective fair value and the classification by level of input within the fair value hierarchy (in millions): 

We classify our cash equivalents and marketable securities within Level 1 or Level 2 because we use quoted market prices or alternative pricing sources and models utilizing market observable inputs to determine their fair value.

We classify our contingent consideration liability within Level 2 as the valuation inputs are based on quoted market prices and market observable data. During the three and six months ended March 31,June 30, 2017, we recognized an increase in the fair value of our contingent consideration liability of $84$21 million and $105 million, respectively, primarily due to the increase in the fair value of our common stock.

Property and equipment consists of the following (in millions): 

Construction in progress includes costs related to construction of data centers, office buildings, and network equipment infrastructure to support our data centers around the world. No interest was capitalized during the three and six months ended March 31,June 30, 2017 and 2016.

The changes in the carrying amount of goodwill for the threesix months ended March 31,June 30, 2017 are as follows (in millions): 

Intangible assets consist of the following (in millions):

Amortization expense of intangible assets was $175$174 million and $180$349 million for the three and six months ended March 31,June 30, 2017, respectively, and $193 million and $373 million for the three and six months ended June 30, 2016, respectively.

As of March 31,June 30, 2017, expected amortization expense for the unamortized acquired intangible assets for the next five years and thereafter is as follows (in millions):

We haveDuring the six months ended June 30, 2017, we entered into additional non-cancelable operating lease agreements, mostly related to office buildings. Our various non-cancelable operating lease agreements for certain of our offices, land, facilities, and data centers withhave original lease periods expiring between 2017 and 2038.2038 and our total future minimum payments related to these operating leases as of June 30, 2017 was $2.7 billion. We are committed to pay a portion of the related actual operating expenses under certain of these lease agreements. Certain of these arrangements have free rent periods or escalating rent payment provisions, and we recognize rent expense under such arrangements on a straight-line basis. Operating lease expense was $74$81 million and $64$156 million for the three and six months ended March 31,June 30, 2017, respectively, and $66 million and $129 million for the three and six months ended June 30, 2016, respectively.

Beginning on May 22, 2012, multiple putative class actions, derivative actions, and individual actions were filed in state and federal courts in the United States and in other jurisdictions against us, our directors, and/or certain of our officers alleging violation of securities laws or breach of fiduciary duties in connection with our initial public offering (IPO) and seeking unspecified damages. We believe these lawsuits are without merit, and we intend to continue to vigorously defend them. The vast majority of the cases in the United States, along with multiple cases filed against The NASDAQ OMX Group, Inc. and The Nasdaq Stock Market LLC (collectively referred to herein as NASDAQ) alleging technical and other trading-related errors by NASDAQ in connection with our IPO, were ordered centralized for coordinated or consolidated pre-trial proceedings in the U.S. District Court for the Southern District of New York. In a series of rulings in 2013 and 2014, the court denied our motion to dismiss the consolidated securities class action and granted our motions to dismiss the derivative actions against our directors and certain of our officers. On July 24, 2015, the court of appeals affirmed the dismissal of the derivative actions. On December 11, 2015, the court granted plaintiffs' motion for class certification in the consolidated securities action. On April 14, 2017, we filed a motion for summary judgment. Trial is scheduled to begin on October 23, 2017.

We are also party to various legal proceedings, claims, and claimsregulatory, tax or government inquiries and investigations that arise in the ordinary course of business. AmongWith respect to these pending matters, one case is currently scheduled for trial inwe evaluate the second quarter of 2017. An entity named Social Ranger LLC sued us for alleged violations of the United States antitrust laws in connection with virtual currency services. The complaint was filed in the

With respect to our other outstanding legal matters, weWe believe that the amount or estimable range of reasonably possible or probable loss will not, either individually or in the aggregate, have a material adverse effect on our business, consolidated financial position, results of operations, or cash flows. However, the outcome of litigationthese matters is inherently uncertain. Therefore, if one or more of these legal matters were resolved against us for amounts in excess of management's expectations, our results of operations and financial condition, including in a particular reporting period in which any such outcome becomes probable and estimable, could be materially adversely affected.

In November 2016, our board of directors authorized a $6.0 billion share repurchase program of our Class A common stock, beginningwhich commenced in 2017 and which does not have an expiration date. The timing and actual number of shares repurchased depend on a variety of factors, including price, general business and market conditions, and other investment opportunities, and shares may be repurchased through open market purchases or privately negotiated transactions, including through the use of trading plans intended to qualify under Rule 10b5-1 under the Exchange Act. During the threesix months ended March 31,June 30, 2017, we repurchased and subsequently retired approximately 23 million shares of our Class A common stock for an aggregate amount of approximately $228$408 million.

We maintain two share-based employee compensation plans: the 2012 Equity Incentive Plan (2012 Plan) and the 2005 Stock Plan (collectively, Stock Plans). Our 2012 Plan serves as the successor to our 2005 Stock Plan and provides for the issuance of incentive and nonstatutory stock options, restricted stock awards, stock appreciation rights, RSUs, performance shares, and stock bonuses to qualified employees, directors and consultants. Outstanding awards under the 2005 Stock Plan continue to be subject to the terms and conditions of the 2005 Stock Plan. Our board of directors approved the amendment and restatement of our 2012 Plan (the Amended 2012 Plan), which was approved by our stockholders and adopted by us in June 2016.

We initially reserved 25 million shares of our Class A common stock for issuance under our 2012 Plan. Following the date of the contemplated stock dividend of Class C capital stock described above, if it is declared and paid, the shares reserved and available for issuance under our Amended 2012 Plan will be shares of the new Class C capital stock, except for shares reserved for awards outstanding immediately prior to the payment of the dividend. The number of shares reserved for issuance under our Amended 2012 Plan increases automatically on January 1 of each of the calendar years during the term of the Amended 2012 Plan, which will continue through and including April 2026 unless terminated earlier by our board of directors or a committee thereof, by a number of shares of Class C capital stock (and prior to the date of the payment of the stock dividend described above, Class A common stock) equal to the lesser of (i) 2.5% of the total issued and outstanding shares of our Class A common stock and

The following table summarizes the activities of stock option awards under the Stock Plans for the threesix months ended March 31,June 30, 2017: 

As of March 31,June 30, 2017, there was $9.48$8.78 billion of unrecognized share-based compensation expense, of which (i) $9.31$8.68 billion was related to RSUs, and (ii) $161$100 million was related to restricted shares, shares related to our contingent consideration with performance conditions that were met in the second quarter of 2016 but are still subject to a service condition, and stock options. This unrecognized compensation expense is expected to be recognized over a weighted-average period of approximately three years.

Our tax provision for interim periods is determined using an estimate of our annual effective tax rate, adjusted for discrete items arising in that quarter. In each quarter, we update our estimate of the annual effective tax rate, and if our estimated annual effective tax rate changes, we make a cumulative adjustment in that quarter. Our quarterly tax provision and our quarterly estimate of our annual effective tax rate are subject to significant volatility due to several factors, including our ability to accurately predict the proportion of our income (loss) before provision for income taxes in multiple jurisdictions, the tax effects of our share-based compensation, and the effects of acquisitions and the integration of those acquisitions.

Our 2017 effective tax rate differs from the U.S. statutory rate primarily due to a portion of our income before provision for income taxes being earned in jurisdictions with tax rates lower than the U.S. statutory rate where we plan to indefinitely reinvest a certain portion of those earnings, as well as the recognition of excess tax benefits from share-based compensation.

Our gross unrecognized tax benefits were $3.37$3.45 billion and $3.31 billion as of March 31,June 30, 2017 and December 31, 2016, respectively. If the gross unrecognized tax benefits as of March 31,June 30, 2017 were realized in a subsequent period, this would result in a tax benefit of $2.64$2.68 billion within our provision of income taxes at such time. Our existing tax positions will continue to

In July 2016, we received a Statutory Notice of Deficiency (Notice) from the IRS related to transfer pricing with our foreign subsidiaries in conjunction with the examination of the 2010 tax year. While the Notice applies only to the 2010 tax year, the IRS states that it will also apply its position to tax years subsequent to 2010, which, if the IRS prevails in its position, could result in

We are subject to taxation in the United States and various other state and foreign jurisdictions. The material jurisdictions in which we are subject to potential examination include the United States and Ireland. We are under examination by the Internal Revenue Service (IRS) for our 2011 through 2013 tax years. Our 2014 and subsequent years remain open to examination by the IRS. Our 2012 and subsequent years remain open to examination in Ireland.

We believe that adequate amounts have been reserved for any adjustments to the provision for income taxes or other tax items that may ultimately result from these examinations. Although the timing of the resolution, settlement, and closure of any audits is highly uncertain, it is reasonably possible that the balance of gross unrecognized tax benefits could significantly change in the next 12 months. Given the number of years remaining that are subject to examination, we are unable to estimate the full range of possible adjustments to the balance of gross unrecognized tax benefits. However, we do not anticipate a significant impact to such amounts within the next 12 months.

Revenue by geography is based on the billing address of the marketer or developer. The following tables set forth revenue and property and equipment, net by geographic area (in millions):

In the firstsecond quarter of 2017, we announced our new mission, which is to give people the power to build community and bring the world closer together. We also continued to focus on our three main revenue growth priorities: (i) helping businesses expand their use of our mobile products, (ii) developing innovative ad products that help businesses get the most of their ad campaigns, and (iii) making our ads more relevant and effective through our targeting capabilities and outcome-based measurement.

We continued to invest, based on our roadmap, in: (i) our most developed ecosystem, the Facebook app and platform, as well as video, (ii) driving growth and building ecosystems around our products and features that already have significant user bases, such as Instagram, Messenger, Instagram, and WhatsApp, and (iii) long-term technology initiatives, such as artificial intelligence, connectivity, and virtual and augmented reality, that we believe will further our mission to connectbuild community and bring the world such as augmented reality, virtual reality and artificial intelligence.closer together. We intend to continue to invest based on this roadmap, and we expect these investments and our increasingly global scale will drive significant overall year-over-year expense growth compared to 2016. In addition, we anticipate our expenses in 2017 will continue to grow as we execute on priorities such as (i) hiring top engineering talent, (ii) investing in research and development, content, and sales and marketing efforts, and (iii) expanding our data center capacity and office facilities to support our rapid growth.

The numbers for our key metrics, our DAUs, MAUs, and average revenue per user (ARPU), do not include Instagram, WhatsApp, or Oculus users unless they would otherwise qualify as such users, respectively, based on their other activities on Facebook. In addition, other user engagement metrics do not include Instagram, WhatsApp, or Oculus unless otherwise specifically stated.

Trends in the number of users affect our revenue and financial results by influencing the number of ads we are able to show, the value of our ads to marketers, the volume of Payments transactions, as well as our expenses and capital expenditures. Substantially all of our daily and monthly active users (as defined below) access Facebook on mobile devices.

During the firstsecond quarter of 2017, worldwide ARPU was $4.23,$4.73, an increase of 27%24% from the firstsecond quarter of 2016. Over this period, ARPU increased by 37%35% in United States & Canada, 36%33% in Europe, 27% in Asia-Pacific, and 40%31% in Rest of World. ARPUWorld, and 20% in the first quarter of 2017 declined by 12% from the fourth quarter of 2016. We believe the sequential quarterly decline was driven by seasonal trends, which also affected ARPU trends from the fourth quarter of 2015 to the first quarter of 2016, during which period ARPU declined by 11%.Asia-Pacific. In addition, user growth was more rapid in geographies with relatively lower ARPU, such as Asia-Pacific and Rest of World. We expect that user growth in the future will be primarily concentrated in those regions where ARPU is relatively lower, such that worldwide ARPU may continue to increase at a slower rate relative to ARPU in any geographic region, or potentially decrease even if ARPU increases in each geographic region.

Share-based compensation expense included in costs and expenses:

Revenue in the second quarter and the first quartersix months of 2017 increased $2.65$2.89 billion, or 49%45%, and $5.54 billion, or 47%, respectively, compared to the same periodperiods in 2016. The increase wasincreases were due to an increaseincreases in advertising revenue.

The most important factor driving advertising revenue growth was an increase in revenue from ads in News Feed. For the second quarter and the first quartersix months of 2017, we estimate that mobile advertising revenue represented approximately 85%87% and 86%, respectively, of total advertising revenue, as compared with approximately 82%84% and 83%, respectively, in the same periodperiods in 2016. Factors that influenced our advertising revenue growth in the second quarter and the first quartersix months of 2017 included (i) an increase in demand for our ad inventory, in part driven by an increase in the number of marketers actively advertising on Facebook, (ii) an increase in user growthusers and their engagement, and (iii) an increase in the number and frequency of ads displayed in News Feed, as well as the quality, relevance, and performance of those ads. However, we anticipate increases in the number and frequency of ads displayed in News Feed will be a less significant driver of our revenue growth in the future.

During the second quarter and first quartersix months of 2017, as compared to the same periodperiods in 2016, the average price per ad increased by 14%24% and 19%, respectively, and the number of ads delivered increased by 32%.19% and 25%, respectively. The increase in average price per ad was driven by an increase in demand for our ad inventory while theinventory. The increase in the ads delivered was driven by the same factors that influenced our advertising revenue growth.growth, partially offset by increasing user engagement with video content and other product changes.

Payments and other fees revenue in the second quarter and the first quartersix months of 2017 decreased $6$40 million, or 3%20%, and $46 million, or 12%, respectively, compared to the same periodperiods in 2016. The decreasedecreases in Payments and other fees revenue waswere mostly due to decreased Payments revenue from games played on personal computers. We anticipate Payments and other fees

The general strengthening of the U.S. dollar relative to certain foreign currencies from the second quarter and the first quartersix months of 2016 compared to the same periodperiods in 2017 had an unfavorable impact on our revenue.

If we had translated revenue for the three monthssecond quarter ended March 31,June 30, 2017 using the prior year's monthly exchange rates for our settlement currencies other than the U.S. dollar, our total revenue and advertising revenue would have been $8.05$9.46 billion and $7.88$9.31 billion, respectively. Using these constant rates, both total revenue and advertising revenue would have been $22$141 million higher than actual total revenue and advertising revenue for the firstsecond quarter of 2017.

Cost of revenue in the second quarter and the first quartersix months of 2017 increased $321$320 million, or 38%35%, and $640 million, or 36%, respectively, compared to the same periodperiods in 2016. The majority of the increase wasincreases in both periods were due to an increaseincreases in operational expenses related to our data centers and technical infrastructure and, to a lesser extent, higher costs associated with ads payment processing and various partnership agreements.arrangements.

Research and development expenses in the second quarter and the first quartersix months of 2017 increased $491$448 million, or 37%30%, and $939 million, or 33%, respectively, compared to the same periodperiods in 2016. The majority of the increase wasincreases in both periods were due to an increaseincreases in payroll and benefits as a result of a 44%48% increase in employee headcount from March 31,June 30, 2016 to March 31,June 30, 2017 in engineering and other technical functions. Additionally, the impact from the change in fair value of our contingent consideration liability andIn addition, our equipment and related expenses to support our research and development efforts increased in both the second quarter and the first quartersix months of 2017, compared to the same periodperiods in 2016.

Marketing and sales expenses in the second quarter and the first quartersix months of 2017 increased $231$223 million, or 28%25%, and $453 million, or 26%, respectively, compared to the same periodperiods in 2016. The majority of the increase wasincreases in both periods in 2017 were partially due to increases in payroll and benefits expenses as a result of a 30% increase in employee headcount from March 31,June 30, 2016 to March 31,June 30, 2017 in our marketing and sales functions,functions. Additionally, in the second quarter and increases inthe first six months of 2017, our marketing expenses increased $61 million and $124 million, respectively, and our professional services expense increased $50 million and $114 million, respectively, due to higher consulting and other professional service fees. Additionally, our marketing expenses increased $63 million in the first quarter of 2017,fees, compared to the same periodperiods in 2016.